GPT-5.3-Codex vs Qwen3.7
Full profiles: GPT-5.3-Codex (OpenAI) and Qwen3.7 (Alibaba): every benchmark, price and rank.
Updated · Data from 9 sources
As of 29 Oct 2026, GPT-5.3-Codex and Qwen3.7 are neck and neck on quality (62.9 vs 60.3 on the BenchLeader Index, inside the ±8.1 margin), and Qwen3.7 is 1.3× cheaper at $3.75 vs $4.81 per 1M tokens blended. Choose GPT-5.3-Codex for agents & tools (80 vs 49), long context (67 vs 64), and coding (61 vs 59); choose Qwen3.7 for fast first answers (14.1 s vs 54.4 s) and instruction following (78 vs 74), where it also streams ~2× faster.
Key facts (as of 29 Oct 2026)
- GPT-5.3-Codex scores 62.9 on the BenchLeader Index and Qwen3.7 scores 60.3 (BenchLeader, 29 Oct 2026).
- GPT-5.3-Codex costs $4.81 per 1M tokens blended (list price, 29 Oct 2026).
- Qwen3.7 costs $3.75 per 1M tokens blended (list price, 29 Oct 2026).
- Qwen3.7 streams ~2× faster than GPT-5.3-Codex (Artificial Analysis, 7-day median, 29 Oct 2026).
- Of the 13 benchmarks both models have, GPT-5.3-Codex leads 9 (BenchLeader, 29 Oct 2026).
GPT-5.3-Codex benchmarks 
Qwen3.7 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 | GPT-5.3-Codex | 61 vs 59 category score; LiveCodeBench 87.3% vs 87.1% | Medium |
| Agents & tool use | GPT-5.3-Codex | 80 vs 49 category score; Terminal-Bench Hard 53.0% vs 50.8% | High |
| Long documents | GPT-5.3-Codex | 67 vs 64 category score; AA-LCR v1.1 83.3% vs 79.0% | Medium |
| Following instructions | Qwen3.7 | 78 vs 74 category score; IFBench 80.5% vs 75.4% | Medium |
| Chat quality (human preference) | – | Only Qwen3.7 has an Arena rating (56 category score) | Not comparable |
| Lowest cost | Qwen3.7 | $3.75 vs $4.81 per 1M blended; agentic coding session $0.180 vs $0.161 | – |
| Fastest first answer | Qwen3.7 | 14.1 s (Qwen3.7 (max)) vs 54.4 s (GPT-5.3-Codex (xhigh)) 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
- GPT-5.3-Codex (xhigh) and Qwen3.7 (max) are neck and neck on quality (62.9 vs 60.3, within the ±8.1 margin).
- GPT-5.3-Codex (xhigh) is stronger in agents & tools, coding, long context, multimodal.
- Qwen3.7 (max) is stronger in instruction following, human preference, knowledge, maths, reasoning.
- Qwen3.7 (max) is 1.3× cheaper ($3.75 vs $4.81 per 1M blended).
- Qwen3.7 (max) streams 2.5× faster (201 vs 80 tokens per second).
At a glance
| Spec | GPT-5.3-Codex | Qwen3.7 |
|---|---|---|
| Developer | OpenAI | Alibaba |
| Released | 5 Feb 2026 | 19 May 2026 |
| Context window (tokens) | 400,000 | 1,000,000 |
| Compared at | GPT-5.3-Codex (xhigh) | Qwen3.7 (max) |
| Open weights | No | No |
| Price in / out / cached input per 1M | $1.75 / $14.00 / $0.175 | $2.50 / $7.50 / $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 | GPT-5.3-Codex (xhigh) | Qwen3.7 (max) | Leader |
|---|---|---|---|
| Reasoning | – | 66.3 | not comparable |
| Coding | 60.9 | 59.1 | GPT-5.3-Codex |
| Agents & tools | 80.0 | 49.3 | GPT-5.3-Codex |
| Maths | – | 58.1 | not comparable |
| Knowledge | – | 61.2 | not comparable |
| Instruction following | 73.5 | 78.0 | Qwen3.7 |
| Human preference | – | 55.6 | not comparable |
| Multimodal | 61.2 | – | not comparable |
| Long context | 66.6 | 64.3 | GPT-5.3-Codex |
Where sources disagree
Across the 13 benchmarks both have, GPT-5.3-Codex leads 9, Qwen3.7 leads 4.
| Overall index | GPT-5.3-Codex (xhigh) | Qwen3.7 (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 62.9 | 60.3 | neck and neck |
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-5.3-Codex (xhigh) | Qwen3.7 (max) |
|---|---|---|
| Blended price per 1M tokens | $4.81 | $3.75 |
| Input / output per 1M | $1.75 / $14.00 | $2.50 / $7.50 |
| Cached input per 1M | $0.175 | $0.500 |
| Price per task: chat reply | $0.0049 | $0.0033 |
| Price per task: summarize a 30-page report | $0.029 | $0.035 |
| Price per task: code edit | $0.032 | $0.026 |
| Price per task: agentic coding session | $0.161 | $0.180 |
| Price per task: structured extraction | $0.0063 | $0.0065 |
| Output speed | 80 tok/s | 201 tok/s |
| Time to first answer | 54.4 s | 14.1 s |
| Time to a complete answer | 60.6 s | 16.6 s |
Each model is compared at its best-ranked reasoning-effort setting. GPT-5.3-Codex (xhigh) scores 62.9 and takes 54 s to its first answer token. Qwen3.7 (max) scores 60.3 and takes 14 s to its first answer token. 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
82 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-5.3-Codex (xhigh) | Qwen3.7 (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 62.9 | 60.3 |
| Agents & tools score | BenchLeader | 80.0 | 49.3 |
| Coding score | BenchLeader | 60.9 | 59.1 |
| Instruction following score | BenchLeader | 73.5 | 78.0 |
| Long context score | BenchLeader | 66.6 | 64.3 |
| Multimodal score | BenchLeader | 61.2 | – |
| Human preference score | BenchLeader | – | 55.6 |
| Knowledge score | BenchLeader | – | 61.2 |
| Maths score | BenchLeader | – | 58.1 |
| Reasoning score | BenchLeader | – | 66.3 |
| Blended price $/M | BenchLeader | $4.81 | $3.75 |
| Output speed | BenchLeader | 80 tok/s | 201 tok/s |
| Time to first answer | BenchLeader | 54.4 s | 14.1 s |
| Context window | BenchLeader | 400,000 tokens | 1,000,000 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | – | 90.9% |
| SimpleBench | SimpleBench | – | 70.4% |
| Arena Hard Prompts | Arena (LMArena) | – | 1495 |
| LiveBench 1.0 Reasoning | LiveBench | – | 83.3% |
| GPQA Diamond (AA) | Artificial Analysis | 91.5% | 92.3% |
| Humanity's Last Exam (AA) | Artificial Analysis | 42.5% | 40.5% |
| GPQA Diamond (Vals) | Vals AI | – | 90.2% |
| CritPt | Artificial Analysis | 16.9% | 13.4% |
| Chess Puzzles | Epoch AI Benchmarking Hub | – | 26.0% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | – | 32.0% |
| LMCA | Epoch AI Benchmarking Hub | – | 44.0% |
| DTBench | Epoch AI Benchmarking Hub | – | 92.3% |
| Coding | |||
| SWE-bench Verified (Epoch) | Epoch AI Benchmarking Hub | 74.8%† | 77.3% |
| SciCode | SciCode | – | 49.5% |
| WeirdML | WeirdML | 77.9% | – |
| Arena Coding | Arena (LMArena) | – | 1523 |
| Arena WebDev | Arena (LMArena) | 1407† | 1515 |
| LiveBench 1.0 Coding | LiveBench | – | 74.2% |
| LiveCodeBench | Vals AI | 87.3% | 87.1% |
| IOI | Vals AI | 53.8% | – |
| SWE-bench (Vals) | Vals AI | 78.0% | 68.8% |
| Code Migration | Vals AI | – | 13.1% |
| Vibe Code Bench v1.1 | Vals AI | 61.8% | 47.7% |
| ALE-Bench | Epoch AI Benchmarking Hub | 1655.2 | 1189.4 |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | – | 1.5% |
Show all 82 benchmarks
| Metric | Publisher | GPT-5.3-Codex (xhigh) | Qwen3.7 (max) |
|---|---|---|---|
| Coding | |||
| Terminal-Bench 2.1 (AA) | Artificial Analysis | – | 74.5% |
| Startrise Frontend | Startrise | – | 63.6% |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 78.4%† | – |
| Arena Agent | Arena (LMArena) | – | -6.6 |
| LiveBench Agentic Coding | LiveBench | – | 43.6% |
| Terminal-Bench Hard | Artificial Analysis | 53.0% | 50.8% |
| τ²-Bench Telecom (AA) | Artificial Analysis | 86.0% | 94.7% |
| Terminal-Bench 2.1 (Vals) | Vals AI | – | 61.0% |
| HiL-Bench | Scale AI SEAL | 4.3%† | – |
| τ³-Banking (AA) | Artificial Analysis | – | 11.8% |
| ITBench SRE (AA) | Artificial Analysis | – | 42.5% |
| Analyst Agent (AA) | Artificial Analysis | – | 18.8% |
| Finance Agent v2 | Vals AI | – | 47.8% |
| Harvey's Legal Agent Benchmark v1 | Vals AI | – | 1.7% |
| Legal Research Bench | Vals AI | – | 25.5% |
| Tax Agent Bench | Vals AI | – | 15.8% |
| Terminal-Bench 2.0 (Vals) | Vals AI | 64.0%† | 59.2% |
| METR Time Horizons | METR | 74.5%† | – |
| GBAEval | Epoch AI Benchmarking Hub | – | 0.4% |
| Vending-Bench 2 | Andon Labs | 5940.1† | – |
| AutomationBench | Artificial Analysis | – | 23.0% |
| GDP.pdf (AA) | Artificial Analysis | – | 8.6% |
| EnterpriseOps-Gym | Artificial Analysis | – | 45.0% |
| AA-Briefcase v1.1 | Artificial Analysis | 868 | 917 |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | – | 64.6% |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | – | 34.1% |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | – | 95.6% |
| ProofBench | Vals AI | – | 26.0% |
| LiveBench Mathematics | LiveBench | – | 85.3% |
| Arena Maths | Arena (LMArena) | – | 1482 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | – | 55.8% |
| LiveBench 1.0 Data Analysis | LiveBench | – | 71.8% |
| MMLU-Pro | Vals AI | – | 89.3% |
| LegalBench | Vals AI | – | 84.9% |
| CorpFin | Vals AI | – | 63.7% |
| TaxEval | Vals AI | – | 75.3% |
| Excel Modeling Benchmark | Vals AI | – | 57.0% |
| MedCode | Vals AI | – | 38.8% |
| MedScribe | Vals AI | – | 79.4% |
| DrugDiscoveryBench | Scale AI SEAL | – | 29.3%† |
| EBR-bench | Epoch AI Benchmarking Hub | – | 9.5% |
| Instruction following | |||
| LiveBench Language | LiveBench | – | 79.7% |
| LiveBench Instruction Following | LiveBench | – | 74.0% |
| IFBench | Artificial Analysis | 75.4% | 80.5% |
| Arena Instruction Following | Arena (LMArena) | – | 1465 |
| Human preference | |||
| Arena Text | Arena (LMArena) | – | 1474 |
| EQ-Bench 4 | EQ-Bench | – | 1100 |
| Arena Creative Writing | Arena (LMArena) | – | 1437 |
| Arena Multi-turn | Arena (LMArena) | – | 1482 |
| Arena Longer Queries | Arena (LMArena) | – | 1490 |
| Multimodal | |||
| MMMU-Pro | Artificial Analysis | 78.5% | – |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 83.3% | 79.0% |
| MLCR | Artificial Analysis | – | 14.4% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 157.1† | 154.1 |
| LiveBench 1.0 | LiveBench | – | 73.1% |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 32.5 | 29.5 |
| AA Progress Index v1 | Artificial Analysis | 301.3 | 293.8 |
Data as of 2026-10-29. Best configuration of each model; every score links to its source on the model pages.
What changed
- GPT-5.3-Codex (xhigh): 1 new result (AA-Briefcase v1.1).
- Qwen3.7 (max): BenchLeader Index 60.8 → 60.3.
- GPT-5.3-Codex (xhigh): 1 new result (AA Progress Index v1).
- Qwen3.7 (max): 1 new result (AA Progress Index v1).
- Qwen3.7 (max): cached input price moved from $0.33/M to $0.50/M (52% up).
- Qwen3.7 (max): cost per task moved from $1.15/M to $0.93/M (19% down).
- Qwen3.7 (max): 1 new result (AA-Briefcase v1.1); 1 score revision; cached input price moved from $0.50/M to $0.33/M (34% down).
- Qwen3.7 (max): cached input price moved from $0.33/M to $0.50/M (52% up).
- Qwen3.7 (max): 1 new result (Harvey's Legal Agent Benchmark v1); cached input price moved from $0.50/M to $0.33/M (34% down).
- Qwen3.7 (max): cached input price moved from $0.33/M to $0.50/M (52% up).
Related comparisons
- GPT-5.3-Codex vs Claude Opus 5.5
- GPT-5.3-Codex vs Gemini 4 Carbon
- GPT-5.3-Codex vs GPT-6 Astra
- Qwen3.7 vs Claude Opus 5.5
- Qwen3.7 vs Gemini 4 Carbon
- Qwen3.7 vs GPT-6 Astra
- Qwen3.7 vs GPT-6.1 Sol
- o3 vs Qwen3.7
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
GPT-5.3-Codex vs Qwen3.7: questions
- Is GPT-5.3-Codex better than Qwen3.7?
- GPT-5.3-Codex (xhigh) and Qwen3.7 (max) are neck and neck on quality (62.9 vs 60.3, within the ±8.1 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 29 Oct 2026, so check the category scores for your use.
- GPT-5.3-Codex vs Qwen3.7: which is cheaper, and will that change?
- Qwen3.7 is cheaper today: $3.75 against $4.81 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is GPT-5.3-Codex or Qwen3.7 better for coding?
- GPT-5.3-Codex scores higher in coding (61 vs 59 on the category index, where 50 is average). On LiveCodeBench, the top shared coding benchmark, it is 87.3% for GPT-5.3-Codex against 87.1% for Qwen3.7.
- Which is faster, GPT-5.3-Codex or Qwen3.7?
- Qwen3.7 streams ~2× faster (201 vs 80 output tokens per second, Artificial Analysis 7-day median).
- Is GPT-5.3-Codex or Qwen3.7 better for agents and tool use?
- GPT-5.3-Codex scores higher in agentic tasks (80 vs 49 on the category index, where 50 is average). On Terminal-Bench Hard, the top shared agents & tools benchmark, it is 53.0% for GPT-5.3-Codex against 50.8% for Qwen3.7.
More questions (4)
- Which is better for long documents?
- GPT-5.3-Codex scores higher on long-context benchmarks (67 vs 64). Qwen3.7 accepts more context: 1,000,000 against 400,000 tokens for GPT-5.3-Codex.
- Which has lower latency?
- Qwen3.7 answers first: 14.1 s against 54.4 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.
- Do independent benchmarks agree on GPT-5.3-Codex vs Qwen3.7?
- Across the 13 benchmarks both have, GPT-5.3-Codex leads 9, Qwen3.7 leads 4.
- When was Qwen3.7 released?
- Qwen3.7 was released on 19 May 2026.