Claude Fable 5 vs o3
Full profiles: Claude Fable 5 (Anthropic) and o3 (OpenAI): every benchmark, price and rank.
Updated · Data from 18 sources
As of 26 Oct 2026, Claude Fable 5 (max) leads on quality (69.3 vs 59.5 on the BenchLeader Index), but o3 costs 5.7× less, so it is the better buy unless you need the extra quality. Pick Claude Fable 5 for knowledge (81 vs 55); pick o3 for agents & tools (70 vs 65).
Claude Fable 5 benchmarks 
o3 benchmarks Which to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Coding | Claude Fable 5 | 74 vs 62 category score | High |
| Agents & tool use | o3 | 70 vs 65 category score; Terminal-Bench Hard 37.1% vs 62.9% | Medium |
| Maths | Claude Fable 5 | 72 vs 71 category score | Medium |
| Reasoning | Claude Fable 5 | 74 vs 52 category score; CritPt 28.6% vs 1.1% | High |
| Knowledge | Claude Fable 5 | 81 vs 55 category score; AA-Omniscience 43.3 vs -15.6 | High |
| Long documents | Claude Fable 5 | 66 vs 62 category score; AA-LCR 82.3% vs 74.7% | Medium |
| Following instructions | o3 | 65 vs 63 category score; IFBench 71.4% vs 63.5% | Medium |
| Chat quality (human preference) | – | Only o3 has an Arena rating (61 category score) | Not comparable |
| Lowest cost | o3 | $3.50 vs $20.00 per 1M blended; agentic coding session $0.152 vs $0.800 | – |
| Fastest first answer | o3 | 6.9 s (o3) vs 90.4 s (Claude Fable 5 (max)) to the first answer token | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- Claude Fable 5 (max) leads on quality: 69.3 vs 59.5.
- Claude Fable 5 (max) is stronger in coding, composite, knowledge, long context, maths, reasoning.
- o3 is stronger in agents & tools, instruction following, human preference, multimodal.
- o3 is 5.7× cheaper ($3.50 vs $20.00 per 1M blended).
- o3 streams 2.5× faster (166 vs 65 tokens per second).
At a glance
| Spec | Claude Fable 5 | o3 |
|---|---|---|
| Developer | Anthropic | OpenAI |
| Released | 9 Jun 2026 | 16 Apr 2025 |
| Context window (tokens) | 1,000,000 | 200,000 |
| Compared at | Claude Fable 5 (max) | o3 |
| Open weights | No | No |
| Price in / out / cached input per 1M | $10.00 / $50.00 / $1.00 | $2.00 / $8.00 / $0.500 |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | Claude Fable 5 (max) | o3 | Leader |
|---|---|---|---|
| Reasoning | 73.8 | 51.9 | Claude Fable 5 |
| Coding | 73.8 | 61.8 | Claude Fable 5 |
| Agents & tools | 65.0 | 70.3 | o3 |
| Maths | 72.4 | 70.6 | Claude Fable 5 |
| Knowledge | 81.0 | 55.3 | Claude Fable 5 |
| Instruction following | 63.1 | 65.1 | o3 |
| Human preference | – | 61.3 | not comparable |
| Multimodal | – | 56.9 | not comparable |
| Long context | 66.3 | 62.3 | Claude Fable 5 |
| Composite | 85.0 | 52.0 | Claude Fable 5 |
Where sources disagree
Across the 11 benchmarks both have, Claude Fable 5 leads 10, o3 leads 1. Every overall index that rates both puts Claude Fable 5 ahead.
| Overall index | Claude Fable 5 (max) | o3 | Leader |
|---|---|---|---|
| BenchLeader Index | 69.3 | 59.5 | Claude Fable 5 |
| AA Intelligence Index v4.3.2 | 49.6 | 20.2 | Claude Fable 5 |
| Arena Text | 1503† | 1432 | Claude Fable 5 |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.
| Measure | Claude Fable 5 (max) | o3 |
|---|---|---|
| Blended price per 1M tokens | $20.00 | $3.50 |
| Input / output per 1M | $10.00 / $50.00 | $2.00 / $8.00 |
| Cached input per 1M | $1.00 | $0.500 |
| Price per task: chat reply | $0.019 | $0.0032 |
| Price per task: summarize a 30-page report | $0.150 | $0.029 |
| Price per task: code edit | $0.135 | $0.024 |
| Price per task: agentic coding session | $0.800 | $0.152 |
| Price per task: structured extraction | $0.030 | $0.0056 |
| Cost to run a full benchmark suite | $8.75 | – |
| Output speed | 65 tok/s | 166 tok/s |
| Time to first answer | 90.4 s | 6.9 s |
| Time to a complete answer | 98.0 s | 9.9 s |
Each model is compared at its best-ranked reasoning-effort setting. Claude Fable 5 (max) scores 69.3 and takes 90 s to its first answer token; Claude Fable 5 scores 66.6 in 4 s, Claude Fable 5 (high) scores 64.4 in 4 s. o3 scores 59.5 and takes 7 s to its first answer token; o3 (high) scores 53.9 in 7 s, o3 (medium) scores 53.3 in 7 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
154 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | Claude Fable 5 (max) | o3 |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 69.3 | 59.5 |
| Agents & tools score | BenchLeader | 65.0 | 70.3 |
| Coding score | BenchLeader | 73.8 | 61.8 |
| Composite score | BenchLeader | 85.0 | 52.0 |
| Instruction following score | BenchLeader | 63.1 | 65.1 |
| Knowledge score | BenchLeader | 81.0 | 55.3 |
| Long context score | BenchLeader | 66.3 | 62.3 |
| Maths score | BenchLeader | 72.4 | 70.6 |
| Reasoning score | BenchLeader | 73.8 | 51.9 |
| Human preference score | BenchLeader | – | 61.3 |
| Multimodal score | BenchLeader | – | 56.9 |
| Blended price $/M | BenchLeader | $20.00 | $3.50 |
| Output speed | BenchLeader | 65 tok/s | 166 tok/s |
| Time to first answer | BenchLeader | 90.4 s | 6.9 s |
| Context window | BenchLeader | 1,000,000 tokens | 200,000 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 85.9% | 81.8%† |
| Humanity's Last Exam | Scale AI / CAIS | – | 20.3%† |
| SimpleBench | SimpleBench | 81.9% | 53.1%† |
| Arena Hard Prompts | Arena (LMArena) | 1529† | 1442 |
| LiveBench 1.0 Reasoning | LiveBench | 89.7% | – |
| GPQA Diamond (AA) | Artificial Analysis | 92.6% | 82.7% |
| Humanity's Last Exam (AA) | Artificial Analysis | 55.5% | 20.1% |
| GPQA Diamond (Vals) | Vals AI | 93.2%† | 84.1%† |
| EnigmaEval | Scale AI SEAL | 39.3%† | – |
| Kagi LLM Benchmark | Kagi LLM Benchmark | 88.8%† | 67.6% |
| GPQA Diamond (HELM) | HELM Capabilities | – | 75.3% |
| ARC-AGI-1 | ARC Prize | 98.5% | 60.8%† |
| ARC-AGI-2 | ARC Prize | 89.2% | 6.5%† |
| CritPt | Artificial Analysis | 28.6% | 1.1% |
| SciPredict | Scale AI SEAL | – | 17.9% |
| Chess Puzzles | Epoch AI Benchmarking Hub | 47.0% | 36.0%† |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 52.0% | 29.0%† |
| CL-bench | Epoch AI Benchmarking Hub | – | 17.8%† |
| LMCA | Epoch AI Benchmarking Hub | 60.3% | 39.7%† |
| DTBench | Epoch AI Benchmarking Hub | 98.4% | 84.8%† |
| ForecastBench | ForecastBench | 66.0% | 64.6% |
| BTF-3 | Epoch AI Benchmarking Hub | 13.0%† | – |
| Coding | |||
| SWE-bench Verified (Epoch) | Epoch AI Benchmarking Hub | – | 62.3%† |
| SciCode | SciCode | 61.0% | – |
| WeirdML | WeirdML | 91.9% | 52.4%† |
Show all 154 benchmarks
| Metric | Publisher | Claude Fable 5 (max) | o3 |
|---|---|---|---|
| Coding | |||
| FrontierCode | Cognition | 53.5%† | – |
| GSO-Bench | GSO-Bench | 78.4%† | 8.8%† |
| Arena Coding | Arena (LMArena) | 1551† | 1460 |
| Arena WebDev | Arena (LMArena) | 1624† | – |
| LiveBench 1.0 Coding | LiveBench | 86.0% | – |
| LiveCodeBench | Vals AI | 89.8%† | 83.9%† |
| SWE-bench (Vals) | Vals AI | 95.0%† | – |
| Aider Polyglot | Aider polyglot leaderboard | – | 81.3% |
| SWE-bench Verified (bash only) | SWE-bench | – | 58.4% |
| SWE-bench Verified (any scaffold) | SWE-bench | – | 58.4% |
| Code Migration | Vals AI | 55.1%† | – |
| ProgramBench | Vals AI | 2.0%† | – |
| Vibe Code Bench v1.1 | Vals AI | 90.3%† | – |
| SWE Atlas: Codebase QnA | Scale AI SEAL | 39.0%† | – |
| SWE Atlas: Refactoring | Scale AI SEAL | 54.8%† | – |
| SWE Atlas: Test Writing | Scale AI SEAL | 55.6%† | – |
| MirrorCode | Epoch AI Benchmarking Hub | 63.9%† | – |
| CadEval | Epoch AI Benchmarking Hub | – | 74.0%† |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | 95.0%† | – |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | 69.7% | – |
| ALE-Bench | Epoch AI Benchmarking Hub | 2041.3† | 933.5† |
| FrontierSWE | Epoch AI Benchmarking Hub | 47.0% | – |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 42.4% | – |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 84.6% | – |
| Startrise Frontend | Startrise | 73.8%† | – |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 44.5% | – |
| GDPval | OpenAI | – | 30.8%† |
| Remote Labor Index | Scale AI / CAIS | 15.8%† | – |
| APEX-Agents | Mercor | 63.6%† | – |
| Arena Agent | Arena (LMArena) | 7.8† | – |
| LiveBench Agentic Coding | LiveBench | 62.2% | – |
| Terminal-Bench Hard | Artificial Analysis | 62.9% | 37.1% |
| τ²-Bench Telecom (AA) | Artificial Analysis | 98.5% | 80.7% |
| Terminal-Bench 2.1 (Vals) | Vals AI | 80.5%† | – |
| MCP Atlas | Scale AI SEAL | 83.3%† | – |
| HiL-Bench | Scale AI SEAL | 56.3%† | – |
| BFCL Overall | Berkeley Function Calling Leaderboard | – | 63.0% |
| τ³-bench Banking | τ²-bench | 39.7% | – |
| τ³-Banking (AA) | Artificial Analysis | 38.1% | – |
| Analyst Agent (AA) | Artificial Analysis | 48.8% | – |
| Finance Agent v2 | Vals AI | 56.3%† | – |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 11.3%† | – |
| Legal Research Bench | Vals AI | 49.5%† | – |
| Tax Agent Bench | Vals AI | 38.3%† | – |
| Terminal-Bench 4.0 (Vals) | Vals AI | 41.4%† | – |
| Terminal-Bench Science | Vals AI | 15.7%† | – |
| Web Search Index | Vals AI | 46.9%† | – |
| Arena Search | Arena (LMArena) | 1229† | – |
| DeepResearch Bench | DeepResearch Bench | – | 45.2%† |
| PostTrainBench | Epoch AI Benchmarking Hub | 41.8% | – |
| METR Time Horizons | METR | – | 63.6% |
| GBAEval | Epoch AI Benchmarking Hub | 74.5%† | – |
| Vending-Bench 2 | Andon Labs | 4966.6 | – |
| AutomationBench | Artificial Analysis | 54.1% | – |
| GDP.pdf (AA) | Artificial Analysis | 24.0% | – |
| EnterpriseOps-Gym | Artificial Analysis | 51.1% | – |
| AA-Briefcase v1.1 | Artificial Analysis | 1537 | – |
| BrowseComp-Plus | BrowseComp-Plus | – | 50.5% |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 87.0% | 33.3%† |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 90.2% | – |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 99.7% | 83.9%† |
| MATH Level 5 | Epoch AI Benchmarking Hub | – | 97.8%† |
| ProofBench | Vals AI | 95.0% | – |
| LiveBench Mathematics | LiveBench | 96.0% | – |
| AIME (Vals) | Vals AI | – | 85.3%† |
| MGSM | Vals AI | – | 91.8%† |
| IMO 2025 | MathArena | – | 16.7%† |
| Omni-MATH (HELM) | HELM Capabilities | – | 71.4% |
| MATH 500 | Vals AI | – | 94.6%† |
| Arena Maths | Arena (LMArena) | 1521† | 1448 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 70.7%† | 49.4%† |
| LiveBench 1.0 Data Analysis | LiveBench | 80.5% | – |
| AA-Omniscience | Artificial Analysis | 43.3 | -15.6 |
| MMLU-Pro | Vals AI | 91.5%† | 85.6%† |
| LegalBench | Vals AI | 88.6%† | 83.8%† |
| CorpFin | Vals AI | 71.8%† | 59.7%† |
| TaxEval | Vals AI | 76.9%† | 74.6%† |
| MedQA | Vals AI | – | 96.1%† |
| PRBench Finance | Scale AI SEAL | 53.9%† | 47.7% |
| PRBench Legal | Scale AI SEAL | 52.6%† | 48.6% |
| MultiNRC | Scale AI SEAL | – | 45.5%† |
| MMLU-Pro (HELM) | HELM Capabilities | – | 85.9% |
| Excel Modeling Benchmark | Vals AI | 73.7%† | – |
| MedCode | Vals AI | 56.1%† | 47.3%† |
| MedScribe | Vals AI | 88.5%† | 76.7%† |
| Public Benefits Bench v1.1 | Vals AI | 70.4%† | – |
| EBR-bench | Epoch AI Benchmarking Hub | 39.5% | – |
| GDP.pdf | Epoch AI Benchmarking Hub | 29.8% | – |
| AA-Omniscience: accuracy | Artificial Analysis | 65.3% | 38.6% |
| AA-Omniscience: non-hallucination | Artificial Analysis | 36.4% | 11.9% |
| Instruction following | |||
| LiveBench Language | LiveBench | 90.7% | – |
| LiveBench Instruction Following | LiveBench | 75.8% | – |
| IFBench | Artificial Analysis | 63.5% | 71.4% |
| MultiChallenge | Scale AI SEAL | – | 56.6%† |
| TutorBench | Scale AI SEAL | – | 52.1%† |
| IFEval (HELM) | HELM Capabilities | – | 86.9% |
| WildBench (HELM) | HELM Capabilities | – | 86.1% |
| Arena Instruction Following | Arena (LMArena) | 1508† | 1403 |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1503† | 1432 |
| EQ-Bench 4 | EQ-Bench | 1310† | – |
| Arena Creative Writing | Arena (LMArena) | 1501† | 1383 |
| Arena Multi-turn | Arena (LMArena) | 1517† | 1420 |
| Arena Longer Queries | Arena (LMArena) | 1522† | 1411 |
| Lech Mazur Creative Writing | Lech Mazur | – | 83.9%† |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1325† | 1214 |
| MMMU-Pro | Artificial Analysis | – | 70.1% |
| VISTA | Scale AI SEAL | – | 50.1%† |
| MMMU (validation) | MMMU | – | 82.9% |
| MMMU-Pro (official) | MMMU | – | 76.4% |
| MMMU-Pro (Vals) | Vals AI | 89.3%† | 80.4%† |
| MortgageTax | Vals AI | 68.9%† | 65.7%† |
| SAGE | Vals AI | 51.9%† | 41.8%† |
| VTB | Scale AI SEAL | – | 13.7% |
| GeoBench | GeoBench | – | 60.0%† |
| VPCT | Epoch AI Benchmarking Hub | – | 52.0%† |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | 38.6%† | – |
| Long context | |||
| Fiction.LiveBench 120k | Fiction.live | – | 100.0%† |
| AA-LCR | Artificial Analysis | 82.3% | 74.7% |
| Arena Document | Arena (LMArena) | 1499† | – |
| MLCR | Artificial Analysis | 20.6% | – |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 163.4† | 146.7 |
| LiveBench 1.0 | LiveBench | 83.0% | – |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 49.6 | 20.2 |
| HELM Capabilities mean | HELM Capabilities | – | 81.1% |
| Vals RSI Index | Vals AI | 24.3% | – |
| Vals Multimodal Index | Vals AI | 74.2%† | – |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | – | 16.0% |
| MASK | Scale AI SEAL | – | 84.5%† |
| PropensityBench | Scale AI SEAL | – | 10.5% |
Data as of 2026-10-26. Best configuration of each model; every score links to its source on the model pages.
What changed
- Claude Fable 5 (max): 1 new result (Vals RSI Index).
- Claude Fable 5 (max): 1 score revision.
- Claude Fable 5 (max): 1 new result (ForecastBench).
- o3: 1 score revision.
- Claude Fable 5 (max): 4 new results (GDPval-AA v2.1, AA-Omniscience: accuracy, AA-Omniscience: non-hallucination…).
- Claude Fable 5 (max): 1 score revision.
- o3: 1 new result (PRBench Legal).
- o3: 1 new result (VTB).
- o3: 1 new result (PRBench Legal).
- o3: 1 new result (PRBench Finance).
Related comparisons
- Claude Fable 5 vs Claude Opus 5.5
- Claude Fable 5 vs Gemini 4 Argon
- Claude Fable 5 vs Muse Wonder Full
- o3 vs Claude Opus 5.5
- o3 vs Gemini 4 Argon
- o3 vs Muse Wonder Full
- o3 vs Claude Fable 5.1
- Claude Fable 5 vs GPT-6 Astra
Rankings: Best LLM overall · Best LLM for coding · Best LLM for agents · Best cheap LLMs
Claude Fable 5 vs o3: questions
- Is Claude Fable 5 better than o3?
- Claude Fable 5 (max) leads on quality: 69.3 vs 59.5. The BenchLeader Index combines every independent quality benchmark; Claude Fable 5 (max) is ahead overall as of 2026-10-26, but check the category scores for your use.
- Which is cheaper, Claude Fable 5 or o3, and will that change?
- o3 is cheaper today: $3.50 against $20.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is Claude Fable 5 better than o3 for coding?
- Claude Fable 5 scores higher in coding (74 vs 62 on the category index, where 50 is average).
- Is Claude Fable 5 better than o3 for agents and tool use?
- o3 scores higher in agentic tasks (70 vs 65 on the category index, where 50 is average). On Terminal-Bench Hard, the top shared agents & tools benchmark, it is 62.9% for Claude Fable 5 against 37.1% for o3.
- Which is better for long documents?
- Claude Fable 5 scores higher on long-context benchmarks (66 vs 62). Claude Fable 5 accepts more context: 1,000,000 against 200,000 tokens for o3.
- Which has lower latency?
- o3 answers first: 6.9 s against 90.4 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
- Which is faster, Claude Fable 5 or o3?
- o3 streams faster: 166 against 65 output tokens per second.
- Do independent benchmarks agree on Claude Fable 5 vs o3?
- Across the 11 benchmarks both have, Claude Fable 5 leads 10, o3 leads 1. Every overall index that rates both puts Claude Fable 5 ahead.
- When was Claude Fable 5 released?
- Claude Fable 5 was released on 9 Jun 2026.