Claude Fable 5 vs GPT-6 Astra
Wins 2 of 8 areas
Coding · Long documents
Wins 4 of 8 areas
Agents · Reasoning · Facts · Math
GPT-6 Astra wins more areas, narrowly.Claude Fable 5 is better at coding.
Both rank among the ten best models we track in reasoning, facts and math.
Scores updated · 41 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking05GPT-6 Astra5 of 6 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own13GPT-6 Astra3 of 4 tests
- MathCompetition and research-level math02GPT-6 Astra2 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up12GPT-6 Astra2 of 3 tests
- CodingWriting and fixing software32Claude Fable 53 of 5 tests
- Long documentsFinding answers in very long texts10Claude Fable 51 of 1 test
- Images and chartsUnderstanding pictures, charts and video11Even1 each
- Following instructionsDoing exactly what it is asked00Even0 each · 1 tie
Long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-6 Astra pulls ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+16.7points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+12.3points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+7.4points ahead
Where Claude Fable 5 pulls ahead
- Programming problems, refreshed regularlyLiveBench · Coding+5.6points ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+4.9points ahead
- Code for real scientific research problemsSciCode+4.5points ahead
Every test, side by side
All 41 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Fable 5
- LMArena · WebDevGPT-6 Astra by 151 rating points16251776+151 rating
- LiveBench · CodingClaude Fable 5 by 5.68680.4+5.6
- LiveBench · Agentic CodingClaude Fable 5 by 4.962.257.3+4.9
- SciCodeClaude Fable 5 by 4.56156.5+4.5
- Terminal-Bench 2.1GPT-6 Astra by 3.884.688.4+3.8
AgentsGPT-6 Astra
- Terminal-Bench 4.0GPT-6 Astra by 16.742.459.1+16.7
- BrowseCompGPT-6 Astra by 4.187.491.5+4.1
- τ-Bench V3 · BankingGPT-6 Astra by 3.338.141.4+3.3
- GDPValClaude Fable 5 by 1.955.553.6+1.9
ReasoningGPT-6 Astra
- ARC-AGI-2GPT-6 Astra by 5.889.295+5.8
- GPQA DiamondGPT-6 Astra by 3.592.696.1+3.5
- CritPtGPT-6 Astra by 3.128.631.7+3.1
- LiveBench · ReasoningGPT-6 Astra by 389.792.7+3
- SimpleBenchGPT-6 Astra by 1.781.983.6+1.7
- Humanity's Last Examtie55.554.7tie
FactsGPT-6 Astra
- AA-Omniscience · Non-hallucinationGPT-6 Astra by 12.336.448.7+12.3
- SimpleQA VerifiedGPT-6 Astra by 4.970.775.6+4.9
- AA-Omniscience · AccuracyClaude Fable 5 by 2.865.362.6+2.8
Images and chartsEven
- LMArena · VisionClaude Fable 5 by 43 rating points13241281+43 rating
- MMMU-ProGPT-6 Astra by 2.784.286.9+2.7
MathGPT-6 Astra
- FrontierMath Tier 4GPT-6 Astra by 7.490.297.6+7.4
- FrontierMath Tiers 1-3 (v2)GPT-6 Astra by 6.78793.7+6.7
- LiveBench · Mathematicstie9696.8tie
Following instructionsEven
- LiveBench · Instruction Followingtie75.875.6tie
Other results16 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- Terminal-Bench-Science 0.1GPT-6 Astra by 39.924.764.6+39.9
- Agents' Last ExamGPT-6 Astra by 33.625.759.3+33.6
- Terminal-Bench 4.0GPT-6 Astra by 13.244.557.7+13.2
- HiL-Bench (Tools-allowed)GPT-6 Astra by 956.365.3+9
- OSWorld 2.0GPT-6 Astra by 6.566.172.6+6.5
- GDPval-AA 2.1Claude Fable 5 by 53 rating points15951542+53 rating
- DeepSWE 1.1GPT-6 Astra by 4.17074.1+4.1
- AA IntelligenceGPT-6 Astra by 3.149.652.7+3.1
- PostTrainBenchGPT-6 Astra by 2.941.444.3+2.9
- HealthBench ProfessionalClaude Fable 5 by 2.66663.4+2.6
- livebench_data_analysisGPT-6 Astra by 2.580.583+2.5
- livebench_languageClaude Fable 5 by 1.390.789.4+1.3
- ARC-AGI-1Claude Fable 5 by 198.597.5+1
- AA Agentic Indextie5151.5tie
- Artificial Analysis Coding Indextie76.576.9tie
- AA-Omnisciencetie43.343.4tie
Questions people ask
Which is better, Claude Fable 5 or GPT-6 Astra?
GPT-6 Astra wins four of the eight areas we test: agents, reasoning, facts and math. Claude Fable 5 wins coding and long documents. They are level on images and charts and following instructions.
Which is better for coding?
Claude Fable 5. It wins 3 of the 5 coding tests both models report; GPT-6 Astra wins 2.
Which is cheaper?
Claude Fable 5 costs $10.00 per million input tokens and $50.00 per million output tokens; GPT-6 Astra costs $10.00 and $50.00. For a million tokens read plus a million written, they cost about the same.
How do you compare the two?
We use the 41 benchmark tests both models have published scores on. The verdict counts the 25 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 16 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.