Claude Fable 5.1 vs Gemini 4 Argon
| Benchmark | Fable 5.1 | Gemini 4 Argon |
|---|---|---|
| Humanity's Last Exam | 53.8 | 57.1 |
| Arena Text · overall | 1,502 · max | 1,524 |
| Arena Text · Chinese | 1,583.6 · max | 1,600.2 |
| Arena Text · English | 1,507.5 · max | 1,525.8 |
| Arena Text · French | 1,525 · max | 1,543.2 |
| Arena Text · instruction following | 1,518.9 · max | 1,527.6 |
| Arena Text · non-English | 1,507.8 · max | 1,518.9 |
| Arena Text · Polish | 1,529.9 · max | 1,530 |
| Arena Text · Russian | 1,518.3 · max | 1,536.8 |
| Arena Text · Spanish | 1,518.8 · max | 1,507.6 |
| AA Omniscience Index | 37.6 | 42.4 |
| SimpleQA Verified v1 | 70.8 · max | 76 |
| CritPt (physics) | 29.1 | 27.1 |
| AA Intelligence Index 4.3.2 | 48.9 | 52.6 |
| AA Omniscience · Accuracy | 63.1 | 49.9 |
| AA Omniscience · Hallucination Rate | 69.1 | 15.1 |
| AA-LCR (long context) 1.1 | 84.7 | 79.7 |
| GDP.pdf · AA all-pass | 26.8 | 21.8 |
| MLCR-AA | 9.4 | 13.9 |
| ARC-AGI-1 · public evaluation | 96.6 | 99.5 |
| ARC-AGI-1 · semi-private | 94.5 | 98.5 |
| ARC-AGI-2 · public evaluation | 87.5 | 97.1 |
| ARC-AGI-2 · semi-private | 86.3 | 91.3 |
| Chess Puzzles | 48 · max | 59 |
| FrontierMath Tier 4 (v2) | 87.8 · max | 78 |
| FrontierMath Tiers 1–3 (v2) | 90.2 · max | 86 |
| Mystery Game Puzzles | 58 · max | 64 |
| MathArena · ArXivMath · August 2026 | 87.7 · high | 82.5 |
| MathArena · BrokenArXiv · August 2026 | 79.8 · high | 54.2 |
| Vals LegalBench | 88.5 | 88.3 |
| Vals MedCode | 53.5 | 58.8 |
| Vals MedScribe | 91.3 | 87.4 |
| Vals Legal Research Bench | 55.3 | 54.8 |
| Arena Text · German | 1,497.8 · max | — |
| Arena Text · Japanese | 1,559.6 · max | — |
| Arena Text · Korean | 1,518 · max | — |
| ARC-AGI-3 · semi-private | — | 49.8 · low |
| FrontierMath Erdos | 0 · max | — |
| MathArena · ArXivMath · April 2026 | 77.5 · max | — |
| MathArena · ArXivMath · June 2026 | 91 · max | — |
| MathArena · ArXivMath · May 2026 | 87.5 · max | — |
| MathArena · ArXivMath · September 2026 | — | 87.7 |
| MathArena · BrokenArXiv · June 2026 | 86.6 · low | — |
| MathArena · BrokenArXiv · May 2026 | 78 · low | — |
| MathArena · BrokenArXiv · September 2026 | — | 49.7 |
| MathArena · ArXivLean · September 2026 | — | 39.6 |
| Vals Science Index | — | 55.4 |
Specs
| Spec | Fable 5.1 | Gemini 4 Argon |
|---|---|---|
| Input, per 1M | $10 | !$2 |
| Output, per 1M | $50 | !$10 |
| Cost per task? | $2.98 | $1.62 |
| Speed, tok/s | 56 | 79 |
| Context | 1M | 1M |
| Weights | Closed | Closed |
| Released | Sep 2026 | Oct 2026 |
| See images | ✓ | ✓ |
| Read PDFs | Partial | ✓ |
| Hear audio | — | ✓ |
| Watch video | — | ✓ |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | Partial |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ✓ |
| Run code | ? | ✓ |
| Use a computer | ? | ? |
What breaks when you switch
Synchronous output stops at 65,536›
With background false, output limits above 65,536 tokens return HTTP 400; stay at or below that limit.
Sampling parameters are ignored›
temperature, top_p and top_k are ignored; remove them from requests.
No explicit caching›
Explicit caching is unavailable; omit explicit cache requests.
Replace forced tool_choice›
Forced tool_choice returns HTTP 400; use auto with an explicit instruction and strict tools, or JSON outputs.
Adaptive thinking is always on›
Disabled thinking or manual budget_tokens returns HTTP 400; omit thinking or use adaptive thinking.
No non-default sampling values›
Non-default temperature, top_p or top_k returns HTTP 400; omit non-default sampling values.
Requires 30-day data retention · Claude API›
A Claude API organization or workspace without 30-day retention receives HTTP 400; configure workspace retention or contact the Anthropic account team.
Thinking blocks bind to conversations›
Where conversation binding is enforced, replaying an invalidated block returns HTTP 400; use the documented drop_block control to continue without it.
No assistant prefill›
Assistant prefill returns HTTP 400; use system prompt instructions instead.