AI Models

AI Models Compared

Every major frontier model, grouped by provider - pricing, context window, benchmark scores, and what each one is actually best at. Continuously web-verified; last checked 2026-11-17.

Recently released

Pricing at a glance

ModelStrong atContextInput / 1MOutput / 1MType
OpenAI (23 models)
GPT-6 CyberNewPreview-1.05M$20.00$100.00Closed
GPT-Image-2.6 FlareNew-Not published$8.00$30.00Closed
GPT-Image-2.6 SunburstNew-Not published$8.00$30.00Closed
GPT-Live-1.1New-Not publishedNot announcedNot announcedClosed
GPT-6.1 TerraNew-1.05M$1.00$5.00Closed
GPT-6 RosalindNewPreview-Not announced$15.00$75.00Closed
GPT-6 LunaNewCost efficiency1.05M$0.10$0.50Closed
GPT-6.1 SolNewLong context1,050,000$2.00$10.00Closed
GPT-6 SolNewLegacyCost efficiency1.05M$2.00$10.00Closed
GPT-Image-2.5 FlareNewLegacy-Not published$8.00$30.00Closed
GPT-6 AstraNewCoding1.05M$10.00$50.00Closed
GPT-Live-1NewLegacySpeedNot published$0.05 per session minuteIncluded (see note)Closed
GPT-5.6-CyberPreviewCoding400K$12.50$75.00Closed
GPT-Realtime-2.1-128K$32.00$64.00Closed
GPT-Transcribe-25 MB audio files$0.0045 per audio minuteNot announcedClosed
GPT-5.6LegacyCoding1.05M$4.00$20.00Closed
GPT-5.6 SolLegacyCoding1.05M$4.00$20.00Closed
GPT-5.5 InstantLegacySpeedNot publishedChatGPT only (no API)ChatGPT only (no API)Closed
GPT-5.5LegacyCoding1,050,000 tokens (128,000 max output)$5.00$30.00Closed
GPT-5.4LegacyLong context1,050,000 tokens (128,000 max output)$2.50$15.00Closed
GPT-5LegacyMath400,000 tokens (128,000 max output)$1.25$10.00Closed
o1LegacyReasoning200,000 tokens (100,000 max output)$15.00$60.00Closed
GPT-4oLegacySpeed128,000 tokens (16,384 max output)$2.50$10.00Closed
Google (15 models)
Gemini 4 HydrogenNew-1M (shared input + output)$1.50$7.50Closed
Gemini Decision 1New-262,144 tokens$0.10N/A (no generated text)Closed
Gemini 4 CarbonNew-1M$2.00$10.00Closed
Gemini 3.8 Flash CyberNewPreview-Not announcedNot announcedNot announcedClosed
Gemini 4 ArgonNewLong context1M (shared input + output)$2.00$10.00Closed
Gemini 3.8 Flash TTSNew-8192 tokens$0.50$9.00Closed
Gemini 3.8 FlashNewLegacyCoding1M$0.75$3.75Closed
Gemini 3.5 Transcribe-Up to 1 hour of audio per request$2.00$12.00Closed
Gemini 3.7 FlashLegacy-1M$0.75$3.75Closed
Gemini 3.5 Flash-Lite-1M$0.30$2.50Closed
Gemini 3.6 FlashLegacySpeed1M$0.75$3.75Closed
Gemini 3.5 Flash CyberPreviewLegacyCodingNot announcedNot announcedNot announcedClosed
Gemini Omni FlashMultimodalNot published$1.50$12.00Closed
Nano Banana 2 LiteSpeed65,536 input / 4,096 output tokens$0.25$30.00Closed
Gemma 4 12BCost efficiency256K tokensFreeFreeOpen
Anthropic (14 models)
Claude Fable 5.2New-1M$8.00$40.00Closed
Claude Mythos 5.2NewPreview-1M$8.00$40.00Closed
Claude Haiku 5.5NewCost efficiency1M$0.10$0.50Closed
Claude Sonnet 5.5NewSpeed1M$2.00$10.00Closed
Claude Opus 5.5NewCoding1M$4.00$20.00Closed
Claude Mythos 5.1NewPreviewLegacyCoding1M$10.00$50.00Closed
Claude Fable 5.1NewLegacy-1M$10.00$50.00Closed
Claude Opus 5LegacyLong context1M$5.00$25.00Closed
Claude Sonnet 5LegacyCoding1M$2.00$10.00Closed
Claude Fable 5LegacyCoding1M$10.00$50.00Closed
Claude Opus 4.8LegacyCoding1M$5.00$25.00Closed
Claude Opus 4.7LegacyCoding1M$5.00$25.00Closed
Claude Sonnet 4.6LegacyLong context1M$3.00$15.00Closed
Claude Haiku 4.5LegacySpeed200K$1.00$5.00Closed
Meta (7 models)
Muse EmberNew-1M$0.10$0.50Closed
Muse Spark OpenNewMultimodal1M$1.25$4.25Open
Muse WonderNewPreviewCoding1M$4.00$20.00Closed
Muse Spark 1.3New-1M$1.25$4.25Closed
Muse GlimmerCost efficiency128KFreeFreeOpen
Muse ImageCost efficiencyNot publishedNot applicable$0.01/imageClosed
Llama 4Long contextUp to 10M tokens (Scout); ~1M tokens (Maverick)FreeFreeOpen
Mistral AI (7 models)
Voxtral 3NewPreviewCost efficiency32K (transcription models)Per audio minute (see note)Per character for TTS (see note)Closed
Mistral Large 4NewCost efficiency1M$0.70$2.10Open
Mistral OCR 4.1-Not published$4.00 per 1,000 pages$5.00 per 1,000 annotated pagesClosed
Robostral NavigatePreviewCost efficiencyN/A (vision-to-action)Contact salesContact salesClosed
Mistral OCR 4LegacyCost efficiency16K$4.00 per 1,000 pages (API) / $2.00 per 1,000 pages (Batch)$5.00 per 1,000 pages (Document AI annotation tier)Closed
Mistral Small 4-256K$0.15$0.60Open
Mistral Large 3LegacyReasoning256K$0.50$1.50Open
Google DeepMind (6 models)
Gemini 3.8 LiveNewSpeed128K$3.00$12.00Closed
Gemini Robotics ER 2PreviewMultimodal128K$1.00$5.00Closed
Gemini 3.5LegacyCoding1,048,576 tokens (Gemini 3.5 Flash; Pro variant not yet released)$1.50$9.00Closed
Gemini 2.5 ProLegacyLong context1,048,576 tokens (1M) input; up to 65K output$1.25$10.00Closed
Gemini 2.5 FlashLegacyCost efficiency1,048,576 tokens (1M) input; up to 65,535 output$0.30$2.50Closed
Gemma 3LegacyCost efficiency128K tokens (32K for the 1B variant)FreeFreeOpen
Alibaba (4 models)
Qwen4-MaxNew-1M$1.50$4.50Closed
Qwen 3.8-Flash-NextCost efficiency262KFreeFreeOpen
Qwen 3.8 27BCoding262K$0.50$3.00Open
Qwen 3.8 MaxLegacyLong context1M$2.00$6.00Closed
Cohere (4 models)
Cohere Parse 5Cost efficiency8K$1.50 per 1,000 pagesN/A (billed per page)Closed
North Mini CodeCost efficiency256K tokensFreeFreeOpen
Command A+-128K$0.30$1.50Open
Command R+LegacyLong context128K tokens$2.50$10.00Closed
DeepSeek (4 models)
DeepSeek V4.1 FlashNew-1M tokens$0.30$1.20Open
DeepSeek V4Long context1M tokens$1.32$3.96Open
DeepSeek V4 FlashLegacyLong context1M tokens$0.30$1.20Open
DeepSeek V3LegacyCost efficiency128K tokensFreeFreeOpen
Z.ai (Zhipu AI) (4 models)
GLM-5.3-1M$1.40$4.40Open
GLM-5.3-FlashLong context1M$0.15$0.50Open
Ox AlphaLegacyLong context1,048,576 (1M tokens)Free during stealth period (ended)Free during stealth period (ended)Closed
GLM-5.2LegacyLong context1M$1.40$4.40Open
Amazon Web Services (2 models)
Amazon Nova 2 Lite-1M$0.30$2.50Closed
Amazon Nova ProMultimodal300K tokens$0.80$3.20Closed
Cognition (2 models)
SWE-2 FlashNew-Not published$0.10$0.50Closed
SWE-2NewCodingNot published$3.00$15.00Closed
Microsoft (2 models)
Microsoft-Decision-1NewPreviewSpeedNot published$0.042N/A (no generated text)Closed
MAI-Cyber-1-FlashPreviewCoding256000Not announcedNot announcedClosed
MiniMax (2 models)
MiniMax H3-ImageNew-Not published$1.50$3.80Open
MiniMax M3.1 FlashNewCost efficiency1M$0.15$0.60Open
Moonshot AI (2 models)
Kimi K3.1 ProNew-1M$2.00$10.00Open
Kimi K3LegacyCoding1M$3.00$15.00Open
NVIDIA (2 models)
Nemotron 3.5 Bolt 74B-A9BNew-256KFreeFreeOpen
Nemotron 3.5 Super VLNewCost efficiency1M$0.10$0.40Open
xAI (SpaceXAI) (2 models)
Grok Voice 3.0New-Not disclosedNot applicableNot applicableClosed
Grok 4.7NewCoding500K$2.00$6.00Closed
Aleph Alpha (1 model)
Kolibri-1NewCost efficiency262K (validated to 1M)FreeFreeOpen
Alibaba (Qwen Team) (1 model)
Qwen 3Cost efficiency128K tokens (32K for 0.6B/1.7B/4B dense variants)FreeFreeOpen
Baidu (1 model)
ERNIE 5.5 Flash PreviewNewPreview-1M$0.28$1.13Closed
Canva (1 model)
Canva CraftNewMultimodalNot publishedNot announcedNot announcedClosed
Cisco (1 model)
Antares-1BCost efficiency128000FreeFreeOpen
Cloudflare (1 model)
ClairicNewPreviewCost efficiency65,536$0.10N/A (no generated text)Closed
Liquid AI (1 model)
LFM3New-256K$0.08$0.20Open
Poolside (1 model)
Laguna S 2.1Coding1M$0.09$0.18Open
Reflection AI (1 model)
BeamNewMath1M$0.25$1.00Open
Salesforce (1 model)
Salesforce KoaNewPreviewReasoningNot publishedNot announcedNot announcedClosed
SpaceXAI (1 model)
Grok 4.6LegacyCost efficiency500K$2.00$6.00Closed
SpaceXAI (xAI) (1 model)
Grok 4.5LegacyCost efficiency500K$2.00$6.00Closed
Springboards (1 model)
FlintPreviewSpeedNot disclosedApp-only (no API)App-only (no API)Closed
StepFun (1 model)
Step 5 PreviewNewPreviewCost efficiency1M$1.00$2.70Open
Tencent (1 model)
Hy4New-1M$0.834$2.501Open
Thomson Reuters (1 model)
ThomsonCost efficiency262000Not sold via public APINot sold via public APIClosed
TypeSafe AI (1 model)
JevNewCost efficiency32K$0.042FreeClosed
Writer (1 model)
Palmyra X6Cost efficiency1000000$2.00$8.00Closed

Prices in USD. Open-source models are free to self-host; API pricing varies by provider.

OpenAI23 models

GPT-6 CyberNew

1.05M ctx

2026-11

OpenAI's cybersecurity variant of GPT-6 Astra, limited to approved Daybreak Red organizations

Approved security teams investigating and reproducing vulnerabilities in authorized test environmentsVerifying that patches fix a vulnerabilityLong-running vulnerability research agents
Full details →

GPT-Image-2.6 FlareNew

Not published ctx

2026-11

OpenAI's fast GPT Image tier for quick iteration, now with GIF and SVG output

Fast, iterative image generation and editingApps where image latency mattersAnimated GIFs and SVG graphics
Full details →

GPT-Image-2.6 SunburstNew

Not published ctx

2026-11

OpenAI's higher-fidelity GPT Image tier for detailed compositions and editing

Detailed marketing and product imagesMulti-step image editing and retouchingFinal renders after drafting with Flare
Full details →

GPT-Live-1.1New

Not published ctx

2026-10

OpenAI's updated full-duplex voice model for ChatGPT, with better accents and 13 new voices

Voice conversations in ChatGPTSpoken practice in the languages this update improvedResearch questions asked and answered by voice
Full details →

GPT-6.1 TerraNew

1.05M ctx

2026-10

OpenAI's fast, low-cost GPT-6 model for computer use, visual reasoning and high-volume work

Computer-use and browser agentsHigh-volume professional work such as document and image analysisWriting and editing where cost and speed matter
Full details →

GPT-6 RosalindNew

Not announced ctx

2026-10

OpenAI's GPT-6 life sciences model, limited to approved research organizations

Life sciences research teams at approved organizationsInternal research that combines biology reasoning with scientific tools
Full details →

GPT-6 LunaNew

1.05M ctx

2026-09

OpenAI's cheapest GPT-6 model, and since October 2026 the default for free ChatGPT users

High-volume, cost-sensitive API workloadsClassification, extraction and routingEveryday chat assistants
Full details →

GPT-6.1 SolNew

1,050,000 ctx

2026-09

Near-Astra intelligence for a fifth of the price

Agentic coding workflows and computer use automationLong-running software engineering tasksDocument-heavy professional work and analysis
Full details →

GPT-6 SolNew

1.05M ctx

2026-09

OpenAI's September 2026 Sol model for coding and agentic work, since replaced by GPT-6.1 Sol

Complex coding and agentic workflowsBusiness workflow automationSoftware engineering and code review
Full details →

GPT-Image-2.5 FlareNew

Not published ctx

2026-09

OpenAI's September 2026 fast image model, since followed by GPT-Image-2.6 Flare at the same price

Fast image generation workflowsProduction-scale image creationIterative creative refinement
Full details →

GPT-6 AstraNew

1.05M ctx

2026-09

OpenAI's most capable model for the most demanding work

Complex reasoning and problem-solvingAgentic workflows and autonomous workScientific research and discovery
Full details →

GPT-Live-1New

Not published ctx

2026-09

A new generation of full-duplex voice models for natural human-AI conversation

Hands-free assistance with cooking, directions, and daily tasksLanguage practice with natural back-and-forth dialogueLive translation during conversations
Full details →

GPT-5.6-Cyber

400K ctx

2026-08

Purpose-trained model for advanced cybersecurity research and vulnerability testing

Vulnerability research and exploitation testingPenetration testing and red-team exercisesExploit validation on authorized systems
Full details →

GPT-Realtime-2.1

128K ctx

2026-07

OpenAI's current speech-to-speech reasoning model for voice agents on the Realtime API

Phone and support voice agentsVoice assistants that call tools mid-conversationSpeech-to-speech apps that need low latency
Full details →

GPT-Transcribe

25 MB audio files ctx

2026-07

OpenAI's recommended speech-to-text model for recorded audio and batch transcription

Meeting, call and podcast transcriptionBatch transcription of recorded audioApps moving off whisper-1 before its February 2027 shutdown
Full details →

GPT-5.6

1.05M ctx

2026-07

OpenAI's GPT-5.6 model family (Sol, Terra, Luna), since followed by GPT-6

Teams that want one generation with multiple cost/quality tiersAgentic coding, knowledge work, and research at scaleWorkloads already tuned to a GPT-5.6 tier
Full details →

GPT-5.6 Sol

1.05M ctx

2026-06

OpenAI's GPT-5.6 flagship tier, now generally available and since replaced by GPT-6 Sol

Frontier agentic coding and computer useComplex multi-step tasks via subagent orchestrationSecurity-sensitive and scientific research workloads
Full details →

GPT-5.5 Instant

Not published ctx

2026-05

OpenAI's fast ChatGPT model from May 2026, no longer the ChatGPT default

Fast, everyday ChatGPT interactionsLatency-sensitive assistant and chat useQuick drafting and summarization
Full details →

GPT-5.5

1,050,000 tokens (128,000 max output) ctx

2026-04

OpenAI's April 2026 frontier model for professional work, since replaced by GPT-5.6 Sol and the GPT-6 family

Agentic coding and multi-tool automationLarge-context document and codebase analysisGeneral-purpose frontier reasoning and knowledge work
Full details →

GPT-5.4

1,050,000 tokens (128,000 max output) ctx

2026-03

OpenAI's March 2026 model for coding and professional work, since superseded by GPT-5.5 and GPT-6

Cost-efficient large-context workGeneral coding and reasoning at scaleProduction workloads migrating within the GPT-5 generation
Full details →

GPT-5

400,000 tokens (128,000 max output) ctx

2025-08

OpenAI's landmark August 2025 flagship: strong reasoning at a low price

Budget-friendly frontier reasoning and codingMath and STEM problem solvingExisting GPT-5-based production systems
Full details →

o1

200,000 tokens (100,000 max output) ctx

2024-12

OpenAI's first-generation deep-reasoning model that thinks before answering

Hard math and science reasoningCompetitive and algorithmic programmingDeliberate multi-step problem solving
Full details →

GPT-4o

128,000 tokens (16,384 max output) ctx

2024-05

OpenAI's versatile, fast multimodal workhorse (text + image)

Fast, low-cost everyday assistant tasksMultimodal (image + text) understandingHigh-volume production workloads
Full details →

Google15 models

Gemini 4 HydrogenNew

1M (shared input + output) ctx

2026-11

Google's fast Gemini 4 model for everyday coding and agent work, with a free API tier

Everyday coding, interactive web apps and tool-using agentsParallel subagent work, including pairing with Gemini 4 Carbon for planningMultimodal document workflows over images, video, audio and PDFs
Full details →

Gemini Decision 1New

262,144 tokens ctx

2026-11

A Google decision model that returns calibrated probabilities for yes/no, choice and rubric questions instead of text

Routing requests between models, tools or agentsClassification, moderation and safety checks on text, images, audio and videoRubric-based grading and LLM-as-judge steps
Full details →

Gemini 4 CarbonNew

1M ctx

2026-10

Google's most capable Gemini model, generally available since November 10, 2026

The hardest long-horizon coding and agentic tasks where Gemini 4 Argon falls shortPlanning and hard roadblocks in a workflow that delegates execution to Gemini 4 HydrogenVulnerability finding, exploitation testing and patching (a more permissive cyber variant is in Google's Fairwind Program)
Full details →

Gemini 3.8 Flash CyberNew

Not announced ctx

2026-09

Google's restricted Flash model for finding and patching software vulnerabilities

Vulnerability discovery across large codebasesAutomated patch generationSecurity teams in government, critical infrastructure and open-source maintenance
Full details →

Gemini 4 ArgonNew

1M (shared input + output) ctx

2026-09

Google's frontier model for real-world coding, enterprise knowledge work, and cybersecurity defense

Long-form software engineering tasksCybersecurity vulnerability analysisEnterprise knowledge work (legal, finance)
Full details →

Gemini 3.8 Flash TTSNew

8192 tokens ctx

2026-09

Advanced text-to-speech with natural language voice design and line-by-line performance direction

Immersive audiobooksGame character voicesPodcast narration
Full details →

Gemini 3.8 FlashNew

1M ctx

2026-09

Google's previous-generation Flash model for coding and agents, at half Gemini 4 Hydrogen's price until December 31, 2026

Software engineering tasksAutonomous agentsComplex enterprise analysis
Full details →

Gemini 3.5 Transcribe

Up to 1 hour of audio per request ctx

2026-08

Google's speech-to-text model that turns raw audio into formatted text

Real-time voice agents and applicationsPost-call analytics and meeting transcriptionMultilingual voice interactions
Full details →

Gemini 3.7 Flash

1M ctx

2026-08

Most intelligent workhorse model yet for coding and agents

Software engineering and code generationAutonomous agent workflows and business automationDocument-heavy knowledge work
Full details →

Gemini 3.5 Flash-Lite

1M ctx

2026-07

Google's fastest, lowest-cost Gemini 3.5-class model for high-volume agent and document work

High-volume classification, extraction and translationSubagents inside larger agent systemsDocument parsing over long PDFs
Full details →

Gemini 3.6 Flash

1M ctx

2026-07

Google's previous-generation Flash model, still served in the Gemini API

Cost-sensitive production AI agentsAgentic workflows requiring efficiencyMulti-step coding tasks
Full details →

Gemini 3.5 Flash Cyber

Not announced ctx

2026-07

Cost-efficient cybersecurity-focused model for vulnerability detection and patching

Automated vulnerability detection at scaleCost-sensitive security scanning workflowsOrganizations needing efficient code security analysis
Full details →

Gemini Omni Flash

Not published ctx

2026-06

Create anything from any input with conversational video editing

Content creators needing iterative video editingBusinesses creating marketing and training videosUsers wanting to generate videos with custom avatars
Full details →

Nano Banana 2 Lite

65,536 input / 4,096 output tokens ctx

2026-06

The fastest, most cost-efficient image generation model in the Nano Banana family

High-volume image generationRapid ideation and A/B testingIterative content creation
Full details →

Gemma 4 12B

256K tokens ctx

2026-06

Open

Google's laptop-runnable open multimodal model with a unified encoder-free design.

Self-hosted multimodal assistantsOn-device / single-GPU deploymentPrivacy-sensitive applications
Full details →

Anthropic14 models

Claude Fable 5.2New

1M ctx

2026-11

Anthropic's top Claude tier for long-horizon agentic work, 20% cheaper than Fable 5.1 at $8/$40

Agent sessions that run for hours, such as large code migrationsMultistep deep research carried through to a finished document, spreadsheet or deckOrchestrating smaller Claude models, or acting as an advisor model they consult
Full details →

Claude Mythos 5.2New

1M ctx

2026-11

Claude Fable 5.2 with fewer safeguards, for organizations in Anthropic's verification programs

Defensive security work at organizations in the Cyber Verification ProgramBiomedical and life sciences research through the Life Sciences Verification ProgramCodebase vulnerability scanning through Claude Security on Claude Enterprise
Full details →

Claude Haiku 5.5New

1M ctx

2026-10

Anthropic's fastest, lowest-cost current model at $0.10 per million input tokens with a 1M context window

High-volume classification, extraction and routingSub-agents and tool-calling loopsCost-sensitive production workloads
Full details →

Claude Sonnet 5.5New

1M ctx

2026-09

Faster, cheaper mid-tier model for everyday tasks and agentic work

Everyday tasks and workflowsAgentic coding and automationDocument, slide, and spreadsheet creation
Full details →

Claude Opus 5.5New

1M ctx

2026-09

Anthropic's current Opus and its recommended starting model, priced at $4/$20 per 1M tokens

Complex coding and agentic tasksLong-horizon reasoningFinancial analysis and compliance work
Full details →

Claude Mythos 5.1New

1M ctx

2026-09

Claude Fable 5.1 with relaxed safeguards for verified organizations, replaced by Mythos 5.2 in November 2026

Defensive security work at organizations in the Cyber Verification ProgramVulnerability research and assessmentLife sciences research through the Life Sciences Verification Program
Full details →

Claude Fable 5.1New

1M ctx

2026-09

Anthropic's September 2026 Fable model, replaced by Fable 5.2 on November 9, 2026 and still in the API at $10/$50

Long-running agentic pipelines where cache-read cost dominatesAutomated scientific research and computational biology workflowsDefensive security work (via the Cyber Verification Program for Mythos 5.1)
Full details →

Claude Opus 5

1M ctx

2026-07

Anthropic's July 2026 Opus for reasoning and agentic work, now replaced by the cheaper Opus 5.5.

coding taskslong-running agentsprofessional knowledge work
Full details →

Claude Sonnet 5

1M ctx

2026-06

Anthropic's June 2026 Sonnet with near-Opus quality, now replaced by Sonnet 5.5 at the same price.

Agentic coding assistants and autonomous dev agentsHigh-volume production agent loops needing strong reasoning per dollarLarge-codebase and long-document analysis with 1M context
Full details →

Claude Fable 5

1M ctx

2026-06

Anthropic's June 2026 Fable model, now legacy; Anthropic recommends migrating to Fable 5.2 ($8/$40)

Long-running autonomous agents on complex, high-value tasksFrontier software engineering and hard multi-service implementationScientific research and advanced analytics
Full details →

Claude Opus 4.8

1M ctx

2026-05

Anthropic's May 2026 Opus for agentic coding and enterprise work, now a legacy model replaced by Opus 5 and Opus 5.5.

Complex, long-horizon agentic coding and autonomous engineeringEnterprise knowledge work and professional-grade analysisHigh-stakes tasks where honesty and flaw-catching matter
Full details →

Claude Opus 4.7

1M ctx

2026-04

The April 2026 Opus with improved high-resolution vision, now a legacy model; Opus 5.5 is the current Opus.

Vision-heavy agentic and document-understanding workloadsTop-tier coding on established Opus 4.7 pipelinesLarge-codebase engineering with 1M context
Full details →

Claude Sonnet 4.6

1M ctx

2026-02

Anthropic's February 2026 Sonnet, now a legacy model; Sonnet 5.5 is the current Sonnet.

Established mid-tier production workloads pinned to a stable modelLarge-context document and codebase analysisAgentic coding at a moderate price point
Full details →

Claude Haiku 4.5

200K ctx

2025-10

Anthropic's October 2025 Haiku, now a legacy model replaced by the cheaper Haiku 5.5.

High-throughput, low-latency production workloadsParallelized sub-agents and multi-agent worker rolesCheap classification, extraction, and summarization at scale
Full details →

Meta7 models

Muse EmberNew

1M ctx

2026-11

Meta's smallest, cheapest Muse model, built for the repeated steps inside agents

Sub-agent work in coding agents: reading files, running tests, targeted fixesHigh-volume tool calling, routing and classificationExtraction across large collections of documents, images or records
Full details →

Muse Spark OpenNew

1M ctx

2026-10

Open

Meta's open-weight release of Muse Spark: an 816B multimodal MoE

Self-hosted multimodal applicationsTeams that want a US-origin open model
Full details →

Muse WonderNew

1M ctx

2026-10

Meta's first frontier model, built for long-running agent tasks

Long-running agent workflowsCoding agentsMultimodal analysis of documents, audio and video
Full details →

Muse Spark 1.3New

1M ctx

2026-09

Meta's proprietary agentic coding model for Muse Code and the Meta Model API

Coding agents and long agentic sessionsQuestions over large codebasesLow-cost bulk work through the Contributor endpoint
Full details →

Muse Glimmer

128K ctx

2026-08

Open

Open-weight agentic AI that runs on consumer GPUs and laptops

Local AI agentsCoding agentsTool use and function calling
Full details →

Muse Image

Not published ctx

2026-07

Image generation model with visual reasoning capabilities built for Meta's ecosystem

Casual Instagram and WhatsApp users generating images in-appAdvertisers creating multiple ad variations automaticallyUsers wanting quick image edits without leaving social apps
Full details →

Llama 4

Up to 10M tokens (Scout); ~1M tokens (Maverick) ctx

2025-04

Open

Meta's natively multimodal open-weight MoE models, with up to 10M tokens of context on Scout.

Extremely long-context open-model applicationsSelf-hosted multimodal deploymentsEfficient MoE inference
Full details →

Mistral AI7 models

Voxtral 3New

32K (transcription models) ctx

2026-10

Mistral's speech-to-text, text-to-speech and voice agent model family

Batch transcription of calls and meetingsRealtime voice agentsEuropean deployments with data residency needs
Full details →

Mistral Large 4New

1M ctx

2026-10

Open

Mistral's 1-trillion-parameter open-weight flagship, nicknamed 'le Chonk'

European organizations with data sovereignty requirementsTeams planning to self-host a frontier-class open modelMultilingual workloads across EU languages
Full details →

Mistral OCR 4.1

Not published ctx

2026-07

Mistral's current OCR model, adding paragraph-level bounding boxes and block confidence scores

PDF and scan to Markdown conversionDocument pipelines that need element positionsExtraction workflows with human review of low-confidence blocks
Full details →

Robostral Navigate

N/A (vision-to-action) ctx

2026-07

8B embodied navigation model for autonomous robot movement using single RGB camera and natural language instructions

Indoor autonomous navigation in offices, warehouses, and commercial buildingsManufacturing, logistics, and delivery applicationsResource-constrained deployments requiring low hardware costs
Full details →

Mistral OCR 4

16K ctx

2026-06

Mistral's June 2026 document OCR model, retired in September 2026 and replaced by OCR 4.1

Document OCR and parsingPDF/scan to structured Markdown or JSONMultilingual document intelligence
Full details →

Mistral Small 4

256K ctx

2026-03

Open

Mistral's open-weight 119B MoE that handles chat, reasoning, image input and coding in one model

Self-hosted general assistantsCost-sensitive API workloadsAgentic coding and tool use on modest hardware
Full details →

Mistral Large 3

256K ctx

2025-12

Open

Mistral Large 3, the previous open-weight Mistral flagship, now deprecated in favor of Mistral Large 4

Existing Mistral Large 3 integrations until the April 30, 2027 retirementSelf-hosted open-weight deployments that do not need a 1T-parameter modelMultilingual reasoning and generation
Full details →

Google DeepMind6 models

Gemini 3.8 LiveNew

128K ctx

2026-09

Most advanced live dialogue model for natural, real-time voice conversations

Production voice agentsReal-time conversational AICustomer service automation
Full details →

Gemini Robotics ER 2

128K ctx

2026-07

High-level reasoning brain for robots enabling video understanding, task orchestration, and multi-robot collaboration

Complex multi-step robotic tasksMulti-robot coordination and collaborationTasks requiring progress monitoring and temporal understanding
Full details →

Gemini 3.5

1,048,576 tokens (Gemini 3.5 Flash; Pro variant not yet released) ctx

2026-05

Google's frontier model for agents and coding, made fast and cheap.

Autonomous coding agentsHigh-volume production apps needing frontier qualityMultimodal document and chart understanding
Full details →

Gemini 2.5 Pro

1,048,576 tokens (1M) input; up to 65K output ctx

2025-06

Google's advanced thinking model for complex reasoning, coding, and long context.

Complex reasoning and STEM problem-solvingLong-context document and codebase analysisHigh-quality multimodal understanding
Full details →

Gemini 2.5 Flash

1,048,576 tokens (1M) input; up to 65,535 output ctx

2025-06

Google's price-performance workhorse with thinking and a 1M-token context.

High-volume production workloadsCost-sensitive chat and RAG appsFast multimodal processing
Full details →

Gemma 3

128K tokens (32K for the 1B variant) ctx

2025-03

Open

Google's previous-generation open, multimodal, multilingual model family.

Local and on-prem multimodal assistantsMultilingual applicationsFine-tuning on custom data
Full details →

Alibaba4 models

Cohere4 models

DeepSeek4 models

Z.ai (Zhipu AI)4 models

Amazon Web Services2 models

Cognition2 models

Microsoft2 models

MiniMax2 models

Moonshot AI2 models

NVIDIA2 models

xAI (SpaceXAI)2 models

Aleph Alpha1 model

Alibaba (Qwen Team)1 model

Baidu1 model

Canva1 model

Cisco1 model

Cloudflare1 model

Liquid AI1 model

Poolside1 model

Reflection AI1 model

Salesforce1 model

SpaceXAI1 model

SpaceXAI (xAI)1 model

Springboards1 model

StepFun1 model

Tencent1 model

Thomson Reuters1 model

TypeSafe AI1 model

Writer1 model

Benchmark scores sourced from official provider release posts. Prices subject to change - check provider pricing pages for current rates.