GPT-5.2-Codex vs Muse Wonder Full
Full profiles: GPT-5.2-Codex (OpenAI) and Muse Wonder Full (Meta): every benchmark, price and rank.
Updated · Data from 5 sources
As of 21 Oct 2026, Muse Wonder Full (max) is the better pick for most workloads: it leads on quality (70.7 vs 59.5 on the BenchLeader Index). Muse Wonder Full leads in every category we measure both in.
GPT-5.2-Codex benchmarks xhigh OpenAI
Muse Wonder Full benchmarks max MetaOpen in the comparison toolWhich 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 |
|---|---|---|---|
| Reasoning | Muse Wonder Full | 95 vs 55 category score; CritPt 32.9% vs 8.7% | High |
| Knowledge | Muse Wonder Full | 71 vs 61 category score; AA-Omniscience 30.1 vs -2.2 | High |
| Long documents | Either | 66 vs 67 category score; AA-LCR 82.3% vs 83.7% | Either |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- Muse Wonder Full (max) leads on quality: 70.7 vs 59.5.
- GPT-5.2-Codex (xhigh) is stronger in agents & tools, instruction following, multimodal.
- Muse Wonder Full (max) is stronger in composite, knowledge, long context, reasoning, coding.
At a glance
| Spec | GPT-5.2-Codex | Muse Wonder Full |
|---|---|---|
| Developer | OpenAI | Meta |
| Released | 18 Dec 2025 | 12 Oct 2026 |
| Context window (tokens) | 400,000 | 1,000,000 |
| Compared at | GPT-5.2-Codex (xhigh) | Muse Wonder Full (max) |
| Open weights | No | No |
| Price in / out / cached input per 1M | $1.75 / $14.00 / $0.175 | – |
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 | GPT-5.2-Codex (xhigh) | Muse Wonder Full (max) | Leader |
|---|---|---|---|
| Reasoning | 55.2 | 95.0 | Muse Wonder Full |
| Coding | – | 72.2 | not comparable |
| Agents & tools | 66.5 | – | not comparable |
| Knowledge | 60.5 | 71.3 | Muse Wonder Full |
| Instruction following | 75.8 | – | not comparable |
| Multimodal | 59.3 | – | not comparable |
| Long context | 66.3 | 67.0 | level |
| Composite | 61.5 | 92.1 | Muse Wonder Full |
Where sources disagree
Across the 6 benchmarks both have, GPT-5.2-Codex leads 0, Muse Wonder Full leads 6. Every overall index that rates both puts Muse Wonder Full ahead.
| Overall index | GPT-5.2-Codex (xhigh) | Muse Wonder Full (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 59.5 | 70.7 | Muse Wonder Full |
| AA Intelligence Index v4.3.2 | 28.5 | 55.8 | Muse Wonder Full |
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 | GPT-5.2-Codex (xhigh) | Muse Wonder Full (max) |
|---|---|---|
| Blended price per 1M tokens | $4.81 | – |
| Input / output per 1M | $1.75 / $14.00 | – / – |
| Cached input per 1M | $0.175 | – |
| Price per task: chat reply | $0.0049 | – |
| Price per task: summarize a 30-page report | $0.029 | – |
| Price per task: code edit | $0.032 | – |
| Price per task: agentic coding session | $0.161 | – |
| Price per task: structured extraction | $0.0063 | – |
| Output speed | 118 tok/s | – |
| Time to first answer | 64.0 s | – |
| Time to a complete answer | 68.3 s | – |
Each model is compared at its best-ranked reasoning-effort setting. GPT-5.2-Codex (xhigh) scores 59.5 and takes 64 s to its first answer token; GPT-5.2-Codex scores 57.3 in 3 s. Muse Wonder Full (max) scores 70.7. 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
58 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 | GPT-5.2-Codex (xhigh) | Muse Wonder Full (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 59.5 | 70.7 |
| Agents & tools score | BenchLeader | 66.5 | – |
| Composite score | BenchLeader | 61.5 | 92.1 |
| Instruction following score | BenchLeader | 75.8 | – |
| Knowledge score | BenchLeader | 60.5 | 71.3 |
| Long context score | BenchLeader | 66.3 | 67.0 |
| Multimodal score | BenchLeader | 59.3 | – |
| Reasoning score | BenchLeader | 55.2 | 95.0 |
| Coding score | BenchLeader | – | 72.2 |
| Blended price $/M | BenchLeader | $4.81 | – |
| Output speed | BenchLeader | 118 tok/s | – |
| Time to first answer | BenchLeader | 64.0 s | – |
| Context window | BenchLeader | 400,000 tokens | 1,000,000 tokens |
| Reasoning | |||
| LiveBench 1.0 Reasoning | LiveBench | 77.7%† | – |
| GPQA Diamond (AA) | Artificial Analysis | 89.9% | – |
| Humanity's Last Exam (AA) | Artificial Analysis | 35.7% | 53.6% |
| CritPt | Artificial Analysis | 8.7% | 32.9% |
| MysteryMechanism | Vals AI | – | 35.1% |
| Coding | |||
| Arena WebDev | Arena (LMArena) | 1339† | 1726 |
| LiveBench 1.0 Coding | LiveBench | 83.6%† | – |
| SciCode (AA) | Artificial Analysis | – | 62.6% |
| LiveCodeBench | Vals AI | 88.0%† | – |
| IOI | Vals AI | – | 79.8% |
| SWE-bench (Vals) | Vals AI | 72.4%† | – |
| SWE-Bench Pro | Scale AI SEAL | 41.0%† | – |
| SWE-bench Verified (bash only) | SWE-bench | 72.8%† | – |
| SWE-bench Verified (any scaffold) | SWE-bench | 72.8%† | – |
| Code Migration | Vals AI | – | 61.0% |
| ProgramBench | Vals AI | – | 0.5% |
| Vibe Code Bench 1-100 | Vals AI | – | 34.0% |
| Vibe Code Bench v1.1 | Vals AI | 37.9%† | 86.0% |
| SWE-Bench Pro (private) | Scale AI SEAL | 41.0%† | – |
| ALE-Bench | Epoch AI Benchmarking Hub | 1299.9† | – |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | – | 57.1% |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 44.9% |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 66.5%† | – |
| LiveBench Agentic Coding | LiveBench | 49.4%† | – |
| Terminal-Bench Hard | Artificial Analysis | 37.1% | – |
Show all 58 benchmarks
| Metric | Publisher | GPT-5.2-Codex (xhigh) | Muse Wonder Full (max) |
|---|---|---|---|
| Agents & tools | |||
| τ²-Bench Telecom (AA) | Artificial Analysis | 92.1% | – |
| GDPval-AA v2.1 | Artificial Analysis | – | 67.8% |
| CyberBench | Vals AI | – | 73.6% |
| Finance Agent v2 | Vals AI | – | 59.1% |
| Harvey's Legal Agent Benchmark | Vals AI | – | 23.3% |
| Legal Research Bench | Vals AI | – | 53.4% |
| SREBench | Vals AI | – | 11.8% |
| Tax Agent Bench | Vals AI | – | 43.0% |
| Terminal-Bench 4.0 (Vals) | Vals AI | – | 58.1% |
| Terminal-Bench Science | Vals AI | – | 37.1% |
| AutomationBench | Artificial Analysis | – | 71.3% |
| GDP.pdf (AA) | Artificial Analysis | – | 32.6% |
| Harvey LAB-AA v1.1 | Harvey | – | 13.1% |
| AA-Briefcase v1.1 | Artificial Analysis | – | 1729 |
| Maths | |||
| LiveBench Mathematics | LiveBench | 88.8%† | – |
| Knowledge | |||
| LiveBench 1.0 Data Analysis | LiveBench | 78.2%† | – |
| AA-Omniscience | Artificial Analysis | -2.2 | 30.1 |
| LegalBench | Vals AI | – | 87.4% |
| BioMysteryBench | Vals AI | – | 73.3% |
| Excel Modeling Benchmark | Vals AI | – | 71.9% |
| MedCode | Vals AI | – | 49.9% |
| MedScribe | Vals AI | – | 89.3% |
| Public Benefits Bench | Vals AI | – | 73.8% |
| AA-Omniscience: accuracy | Artificial Analysis | 41.1% | 43.2% |
| AA-Omniscience: non-hallucination | Artificial Analysis | 26.6% | 77.0% |
| Instruction following | |||
| LiveBench Language | LiveBench | 73.7%† | – |
| LiveBench Instruction Following | LiveBench | 66.5%† | – |
| IFBench | Artificial Analysis | 77.6% | – |
| Multimodal | |||
| MMMU-Pro | Artificial Analysis | 76.3% | – |
| Long context | |||
| AA-LCR | Artificial Analysis | 82.3% | 83.7% |
| Composite | |||
| LiveBench 1.0 | LiveBench | 74.0%† | – |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 28.5 | 55.8 |
| Vals Index | Vals AI | – | 66.8 |
Data as of 2026-10-21. Best configuration of each model; every score links to its source on the model pages.
What changed
- Muse Wonder Full (max): 1 new result (FrontierCode 1.1 (AA)).
- Muse Wonder Full (max): BenchLeader Index 68.6 → 70.7.
- Muse Wonder Full (max): 1 new result (Arena WebDev); 2 score revisions.
- Muse Wonder Full (max): 21 new results (IOI, LegalBench, Vals Index…).
- Muse Wonder Full (max): 1 new result (Harvey LAB-AA v1.1).
- Muse Wonder Full (max): first results.
- GPT-5.2-Codex (xhigh): BenchLeader Index 60.0 → 59.5.
- GPT-5.2-Codex (xhigh): BenchLeader Index 60.7 → 60.0.
- GPT-5.2-Codex (xhigh): 2 new results (AA-Omniscience: accuracy, AA-Omniscience: non-hallucination).
Related comparisons
- GPT-5.2-Codex vs Claude Opus 5.5
- GPT-5.2-Codex vs Gemini 4 Argon
- GPT-5.2-Codex vs GPT-6 Astra
- Muse Wonder Full vs Claude Opus 5.5
- Muse Wonder Full vs Gemini 4 Argon
- Muse Wonder Full vs GPT-6 Astra
- Muse Wonder Full vs Claude Fable 5.1
- GPT-5.2 vs Muse Wonder Full
Rankings: Best LLM overall · Best LLM for coding · Best LLM for agents · Best cheap LLMs
GPT-5.2-Codex vs Muse Wonder Full: questions
- Is GPT-5.2-Codex better than Muse Wonder Full?
- Muse Wonder Full (max) leads on quality: 70.7 vs 59.5. The BenchLeader Index combines every independent quality benchmark; Muse Wonder Full (max) is ahead overall as of 2026-10-21, but check the category scores for your use.
- Which is better for long documents?
- They are level on long-context benchmarks (66 vs 67). Muse Wonder Full accepts more context: 1,000,000 against 400,000 tokens for GPT-5.2-Codex.
- Do independent benchmarks agree on GPT-5.2-Codex vs Muse Wonder Full?
- Across the 6 benchmarks both have, GPT-5.2-Codex leads 0, Muse Wonder Full leads 6. Every overall index that rates both puts Muse Wonder Full ahead.
- When was Muse Wonder Full released?
- Muse Wonder Full was released on 12 Oct 2026.