For Developers/Models/Compare/Claude Fable 5.2 vs GPT-6.1 Sol

Claude Fable 5.2 vs GPT-6.1 Sol

Pricing, benchmarks, and use case comparison

Verdict

Pick Claude Fable 5.2 for the hardest long-horizon and scientific agent work, where it leads Terminal-Bench-Science 0.1 by 17.6 points at max effort; pick GPT-6.1 Sol for agentic coding and high-volume work, where DeepSWE results are close and it costs a quarter as much per token ($2/$10 vs $8/$40).

Specs comparison

Claude Fable 5.2GPT-6.1 Sol
ProviderAnthropicOpenAI
TypeClosed sourceClosed source
Context window✓1M1,050,000
Input / 1M tokens$8.00✓$2.00
Output / 1M tokens$40.00$10.00
Release date2026-112026-09

Benchmarks

BenchmarkClaude Fable 5.2GPT-6.1 Sol
SWE-bench Pro93.0% (max effort)-
Terminal-Bench 4.067.6% (max effort)-
DeepSWE v1.174.3% (max effort)✓75.2% (high effort)
FrontierCode 2.049.1% (xhigh effort)-
Terminal-Bench-Science 0.174.6%-
Humanity's Last Exam (with tools)74.2%-
OSWorld 2.189.3%-
OSWorld 2.0 offline subset-71.4% (max effort)
Terminal-Bench Science 0.1 (Max Effort)-Cost: $5.47/task
AutomationBench (Medium Effort)-2.2 points above Claude Opus 5.5
Factual Accuracy (Low Reasoning Effort)-7.7% error rate

Scores sourced from official provider release posts and independent benchmark aggregators.

Capability and benchmarks

Two benchmarks have vendor-reported scores for both models at max effort. On DeepSWE v1.1, Claude Fable 5.2 averages 74.3% over five trials and GPT-6.1 Sol scores 71.9%. Sol does better at high effort, where it reaches 75.2%, so on this coding test the two are close. Terminal-Bench-Science 0.1, which tests agents on scientific research workflows, shows a wide gap: Fable 5.2 scores 74.6% against 57.0% for Sol, and is also above the 68.1% OpenAI reports for its GPT-6 Astra flagship. Anthropic also reports 93.0% on SWE-bench Pro and 67.6% on Terminal-Bench 4.0 for Fable 5.2; OpenAI has not published either benchmark for Sol. Each lab ran its own model in its own harness, so read small gaps with caution.

Fable 5.2 ships with safety classifiers that block high-risk cybersecurity work, such as exploit generation and penetration testing, and most research biology tasks; Anthropic answers blocked requests with an older Claude model. Organizations verified through Anthropic's programs can use Claude Mythos 5.2, the same model with more permissive safeguards.

Price and context

Claude Fable 5.2 costs $8 input / $40 output per 1M tokens, 20% below Fable 5.1, with cache reads at $0.20. GPT-6.1 Sol costs $2/$10 with cached input at $0.10, a quarter of Fable's per-token price. Sol's discount narrows on long prompts: a request above 272K input tokens pays double for input and 1.5x for output, while Fable 5.2 bills its full 1M-token context at the standard rate. Sol's window is 1,050,000 tokens (922K max input), and both cap output at 128K tokens. Defaults differ too: Fable 5.2 always thinks and defaults to high effort on the API, while Sol defaults to medium. Fable 5.2 offers a fast mode at $16/$80 for up to 2.5x faster output, and it requires 30-day data retention, so it is not available under zero data retention unless Anthropic authorizes it.

  • Claude Fable 5.2: $8/$40 across the full 1M window.
  • GPT-6.1 Sol: $2/$10 up to 272K input tokens, then $4/$15 for the whole request.

Which to pick

  • Pick Claude Fable 5.2 for long-running agent sessions, scientific and research workflows, and very long prompts, where its Terminal-Bench-Science lead and flat 1M-context pricing justify the higher price.
  • Pick GPT-6.1 Sol for agentic coding and repeated, cost-sensitive workloads, where its DeepSWE results come close to Fable 5.2's at a quarter of the per-token price, and if your stack already runs on the OpenAI API or Codex.
  • For a middle option, Claude Opus 5.5 vs GPT-6.1 Sol compares Anthropic's $4/$20 recommended starting model with Sol.

Which should you choose?

Choose Claude Fable 5.2 if...

  • →Long-running agentic coding or research where Opus 5.5 at higher effort still falls short
  • →You run Fable 5.1 today: 5.2 costs less ($8/$40 vs $10/$50) and scores higher on Anthropic's coding and agent benchmarks
  • →You want one strong model to orchestrate or advise cheaper Claude models
Full Claude Fable 5.2 details →

Choose GPT-6.1 Sol if...

  • →When you need Astra-level intelligence but have a 1/5 budget constraint
  • →For repeated and long-running workloads where cost per task matters
  • →When coding performance and computer use are primary requirements
Full GPT-6.1 Sol details →

Compare Claude Fable 5.2 with others