Claude Opus 4.6 vs GLM 5.2
Full profiles: Claude Opus 4.6 (Anthropic) and GLM 5.2 (Zhipu AI): every benchmark, price and rank.
Updated · Data from 9 sources
As of 22 Nov 2026, Claude Opus 4.6 and GLM 5.2 are neck and neck on quality (62.4 vs 58.8 on the BenchLeader Index, inside the ±4.9 margin), and GLM 5.2 is 4.7× cheaper at $2.15 vs $10.00 per 1M tokens blended. Claude Opus 4.6 leads in every category we measure both in. GLM 5.2 streams ~1.5× faster.
Key facts (as of 22 Nov 2026)
- Claude Opus 4.6 scores 62.4 on the BenchLeader Index and GLM 5.2 scores 58.8 (BenchLeader, 22 Nov 2026).
- Claude Opus 4.6 costs $10.00 per 1M tokens blended (list price, 22 Nov 2026).
- GLM 5.2 costs $2.15 per 1M tokens blended (list price, 22 Nov 2026).
- GLM 5.2 streams ~1.5× faster than Claude Opus 4.6 (Artificial Analysis, 7-day median, 22 Nov 2026).
- Of the 14 benchmarks both models have, Claude Opus 4.6 leads 13 (BenchLeader, 22 Nov 2026).
- On Arena WebDev, Claude Opus 4.6 is rated 1546 and GLM 5.2 1608 (LMArena, as published by 22 Nov 2026).
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 | Either | 62 vs 62 category score; Arena WebDev 1546 vs 1608 | Either |
| Maths | Claude Opus 4.6 | 63 vs 53 category score; AIME 2026 96.7% vs 90.0% | High |
| Reasoning | Claude Opus 4.6 | 64 vs 57 category score; Arena Hard Prompts 1530 vs 1496 | Medium |
| Chat quality (human preference) | Claude Opus 4.6 | 70 vs 66 on Arena votes | Medium |
| Lowest cost | GLM 5.2 | $2.15 vs $10.00 per 1M blended; agentic coding session $0.102 vs $0.400 | – |
| Fastest first answer | Claude Opus 4.6 | 1.4 s (Claude Opus 4.6 (high)) vs 34.1 s (GLM 5.2 (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 Opus 4.6 (high) and GLM 5.2 (max) are neck and neck on quality (62.4 vs 58.8, within the ±4.9 margin).
- Claude Opus 4.6 (high) is stronger in coding, human preference, maths, reasoning.
- GLM 5.2 (max) is 4.7× cheaper ($2.15 vs $10.00 per 1M blended).
- GLM 5.2 (max) streams 1.6× faster (58 vs 37 tokens per second).
At a glance
| Spec | Claude Opus 4.6 | GLM 5.2 |
|---|---|---|
| Developer | Anthropic | Zhipu AI |
| Released | 5 Feb 2026 | 16 Jun 2026 |
| Context window (tokens) | 1,000,000 | 1,000,000 |
| Compared at | Claude Opus 4.6 (high) | GLM 5.2 (max) |
| Open weights | No | Yes |
| Price in / out / cached input per 1M | $5.00 / $25.00 / $0.500 | $1.40 / $4.40 / $0.235 |
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 Opus 4.6 (high) | GLM 5.2 (max) | Leader |
|---|---|---|---|
| Reasoning | 64.2 | 56.8 | Claude Opus 4.6 |
| Coding | 62.0 | 61.5 | level |
| Agents & tools | – | 61.7 | not comparable |
| Maths | 63.4 | 52.5 | Claude Opus 4.6 |
| Knowledge | – | 44.2 | not comparable |
| Instruction following | – | 71.9 | not comparable |
| Human preference | 69.7 | 66.3 | Claude Opus 4.6 |
| Multimodal | 66.9 | – | not comparable |
| Long context | – | 63.8 | not comparable |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
Not drawn, measured for fewer than two of these models: Agents & tools, Knowledge, Instruction following, Long context, Multimodal.
Where sources disagree
Across the 14 benchmarks both have, Claude Opus 4.6 leads 13, GLM 5.2 leads 1. Every overall index that rates both puts Claude Opus 4.6 ahead.
| Overall index | Claude Opus 4.6 (high) | GLM 5.2 (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 62.4 | 58.8 | neck and neck |
| Arena Text | 1503 | 1475 | Claude Opus 4.6 |
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. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | Claude Opus 4.6 (high) | GLM 5.2 (max) |
|---|---|---|
| Blended price per 1M tokens | $10.00 | $2.15 |
| Input / output per 1M | $5.00 / $25.00 | $1.40 / $4.40 |
| Cached input per 1M | $0.500 | $0.235 |
| Price per task: chat reply | $0.0095 | $0.0019 |
| Price per task: summarize a 30-page report | $0.075 | $0.019 |
| Price per task: code edit | $0.068 | $0.015 |
| Price per task: agentic coding session | $0.400 | $0.102 |
| Price per task: structured extraction | $0.015 | $0.0037 |
| Agentic coding session, input mostly cached | $0.198 | $0.049 |
| Output speed | 37 tok/s | 58 tok/s |
| Time to first answer | 1.4 s | 34.1 s |
| Time to a complete answer, last measured 13 Nov 2026 | – | 56.7 s |
Each model is compared at its best-ranked reasoning-effort setting. Claude Opus 4.6 (high) scores 62.4 and takes 1 s to its first answer token; Claude Opus 4.6 scores 60.7 in 1 s, Claude Opus 4.6 (no reasoning) scores 57.1 in 2 s. GLM 5.2 (max) scores 58.8 and takes 34 s to its first answer token; GLM 5.2 scores 51.8 in 1 s, GLM 5.2 (no reasoning) scores 45.7 in 2 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
153 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 Opus 4.6 (high) | GLM 5.2 (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 62.4 | 58.8 |
| Coding score | BenchLeader | 62.0 | 61.5 |
| Human preference score | BenchLeader | 69.7 | 66.3 |
| Maths score | BenchLeader | 63.4 | 52.5 |
| Multimodal score | BenchLeader | 66.9 | – |
| Reasoning score | BenchLeader | 64.2 | 56.8 |
| Agents & tools score | BenchLeader | – | 61.7 |
| Instruction following score | BenchLeader | – | 71.9 |
| Knowledge score | BenchLeader | – | 44.2 |
| Long context score | BenchLeader | – | 63.8 |
| Blended price $/M | BenchLeader | $10.00 | $2.15 |
| Output speed | BenchLeader | 37 tok/s | 58 tok/s |
| Time to first answer | BenchLeader | 1.4 s | 34.1 s |
| Context window | BenchLeader | 1,000,000 tokens | 1,000,000 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 90.5%† | 91.9% |
| Humanity's Last Exam | Scale AI / CAIS | 19.0%† | – |
| SimpleBench | SimpleBench | 67.6%† | 58.8%† |
| Arena Hard Prompts | Arena (LMArena) | 1530 | 1496 |
| LiveBench 1.0 Reasoning | LiveBench | 88.7% | 78.6%† |
| GPQA Diamond (AA) | Artificial Analysis | 84.0%† | 89.5% |
| Humanity's Last Exam (AA) | Artificial Analysis | 19.1%† | 41.1% |
| HLE-Diamond (AA) | Artificial Analysis | – | 17.7% |
| GPQA Diamond (Vals) | Vals AI | 89.7%† | 85.6%† |
| Kagi LLM Benchmark | Kagi LLM Benchmark | 72.4%† | 62.6%† |
| ARC-AGI-1 | ARC Prize | 94.0% | 77.0%† |
| ARC-AGI-2 | ARC Prize | 69.2% | 22.8%† |
| ARC-AGI-3 | ARC Prize | 0.5%† | – |
| CritPt | Artificial Analysis | 2.8%† | 20.9% |
| HLE Text-Only (Scale) | Scale AI SEAL | 19.4%† | – |
| Arena Expert | Arena (LMArena) | 1544 | 1503 |
| Chess Puzzles | Epoch AI Benchmarking Hub | 21.0%† | 22.0% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 15.0%† | 18.0%† |
| CL-bench | Epoch AI Benchmarking Hub | 20.7%† | – |
| CL-bench Life | Epoch AI Benchmarking Hub | 17.0% | – |
| LMCA | Epoch AI Benchmarking Hub | 55.8%† | 45.8% |
| DTBench | Epoch AI Benchmarking Hub | 91.2%† | 93.6% |
| ForecastBench | ForecastBench | 63.1%† | – |
| Coding | |||
| SWE-bench Verified (Epoch) | Epoch AI Benchmarking Hub | 78.7%† | 78.7% |
| SciCode | SciCode | – | 51.2% |
Show all 153 benchmarks
| Metric | Publisher | Claude Opus 4.6 (high) | GLM 5.2 (max) |
|---|---|---|---|
| Coding | |||
| WeirdML | WeirdML | 78.0% | 70.1% |
| FrontierCode | Cognition | 26.6%† | 24.5%† |
| GSO-Bench | GSO-Bench | 41.2% | – |
| Arena Coding | Arena (LMArena) | 1547 | 1516 |
| Arena WebDev | Arena (LMArena) | 1546 | 1608 |
| LiveBench 1.0 Coding | LiveBench | 78.2% | 79.7%† |
| LiveCodeBench | Vals AI | 84.7%† | 69.5%† |
| SWE-bench (Vals) | Vals AI | 78.2%† | 82.8% |
| SWE-Bench Pro | Scale AI SEAL | 51.9%† | – |
| SWE-bench Verified (bash only) | SWE-bench | 75.6%† | – |
| SWE-bench Multilingual (bash only) | SWE-bench | 72.0%† | – |
| SWE-bench Verified (any scaffold) | SWE-bench | 75.6%† | – |
| Code Migration | Vals AI | – | 37.9% |
| ProgramBench | Vals AI | – | 0.5% |
| Vibe Code Bench v1.1 | Vals AI | 57.6%† | 64.0% |
| SWE-Bench Pro (private) | Scale AI SEAL | 47.1%† | – |
| SWE Atlas v1.0: Codebase QnA | Scale AI SEAL | 33.3%† | 48.1%† |
| SWE Atlas v1.0: Refactoring | Scale AI SEAL | 35.6%† | 42.4%† |
| SWE Atlas v1.0: Test Writing | Scale AI SEAL | 36.7%† | 41.5%† |
| MirrorCode | Epoch AI Benchmarking Hub | 33.3%† | 36.7% |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | – | 55.6%† |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | – | 43.8% |
| ALE-Bench | Epoch AI Benchmarking Hub | 996.5† | 1010.2 |
| AlgoTune | AlgoTune | 1.5 | – |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | – | 1.0% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | – | 77.9% |
| Startrise Frontend | Startrise | – | 72.1%† |
| GameBench 2 Overall | DuelLab GameBench 2 | – | 43.1† |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | – | 32.4† |
| Design Arena Code | Design Arena | 1271† | 1268† |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 79.8%† | – |
| Cybench | Cybench | 93.0%† | – |
| Remote Labor Index | Scale AI / CAIS | 4.2%† | – |
| APEX-Agents | Mercor | 46.3%† | 45.2%† |
| Arena Agent | Arena (LMArena) | – | -0.4 |
| LiveBench Agentic Coding | LiveBench | 49.0% | 51.8%† |
| Terminal-Bench Hard | Artificial Analysis | 48.5%† | 50.8% |
| τ²-Bench Telecom (AA) | Artificial Analysis | 84.8%† | 99.1% |
| Terminal-Bench 2.1 (Vals) | Vals AI | – | 67.8% |
| MCP Atlas | Scale AI SEAL | 76.8%† | 77.8%† |
| HiL-Bench | Scale AI SEAL | 38.3%† | 43.7%† |
| τ³-bench Banking | τ²-bench | 27.3%† | 37.1%† |
| GDPval-AA v2.1 | Artificial Analysis | – | 1380 |
| τ³-Banking (AA) | Artificial Analysis | – | 34.6% |
| ITBench SRE (AA) | Artificial Analysis | – | 42.7% |
| APEX-Agents (AA) | Artificial Analysis | 33.0%† | 33.7% |
| Finance Agent v2 | Vals AI | – | 49.7%† |
| Harvey's Legal Agent Benchmark v1 | Vals AI | – | 7.1% |
| Legal Research Bench | Vals AI | – | 31.3% |
| SkillsBench | Vals AI | – | 45.1% |
| SREBench | Vals AI | – | 0.0% |
| Terminal-Bench 2.0 (Vals) | Vals AI | 58.4%† | – |
| Arena Search | Arena (LMArena) | 1222† | – |
| Arena Agent Task Completion | Arena (LMArena) | – | 1.2 |
| Arena Search Factuality | Arena (LMArena) | 1227† | – |
| DeepResearch Bench | DeepResearch Bench | 55.3% | – |
| PostTrainBench | Epoch AI Benchmarking Hub | – | 30.4% |
| METR Time Horizons | METR | 78.9%† | – |
| METR 50% Time Horizon | METR | 11 h 59 min† | – |
| GBAEval | Epoch AI Benchmarking Hub | 44.1%† | 0.0%† |
| Vending-Bench 2 | Andon Labs | 8017.6† | 8313.8† |
| AutomationBench | Artificial Analysis | – | 28.4% |
| GDP.pdf (AA) | Artificial Analysis | – | 10.4% |
| EnterpriseOps-Gym | Artificial Analysis | – | 42.7% |
| AA-Briefcase v1.1 | Artificial Analysis | – | 1226 |
| Design Arena Agentic | Design Arena | 1186† | 1184† |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 66.0%† | 59.2% |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 26.8%† | 29.3% |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 94.4%† | 86.4% |
| ProofBench | Vals AI | 50.0%† | 35.0% |
| LiveBench Mathematics | LiveBench | 89.3% | 89.8%† |
| AIME (Vals) | Vals AI | 95.6%† | – |
| AIME 2026 | MathArena | 96.7% | 90.0% |
| HMMT February 2026 | MathArena | 96.2% | 92.4% |
| USAMO 2026 | MathArena | 47.0% | – |
| MathArena Apex | MathArena | 34.5% | 19.8% |
| Arena Maths | Arena (LMArena) | 1516 | 1488 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 47.0%† | 34.2% |
| LiveBench 1.0 Data Analysis | LiveBench | 69.9% | 73.7%† |
| AA-Omniscience v1.1 | Artificial Analysis | – | 0.6 |
| MMLU-Pro | Vals AI | 89.1%† | 86.7%† |
| LegalBench | Vals AI | 85.3%† | 84.1%† |
| CorpFin | Vals AI | 67.0%† | 66.1%† |
| TaxEval | Vals AI | 76.0%† | 73.3%† |
| MedQA | Vals AI | 95.4%† | – |
| PRBench Finance | Scale AI SEAL | 53.3%† | – |
| PRBench Legal | Scale AI SEAL | 52.3%† | – |
| MultiNRC | Scale AI SEAL | 48.3%† | – |
| CaseLaw v2 | Vals AI | 62.1%† | – |
| Excel Modeling Benchmark | Vals AI | – | 61.5%† |
| MedCode | Vals AI | 49.1%† | 40.8%† |
| MedScribe | Vals AI | 86.7%† | 83.5%† |
| DrugDiscoveryBench | Scale AI SEAL | 27.7%† | 36.2%† |
| Arena Text Factuality | Arena (LMArena) | 1493 | 1463 |
| EBR-bench | Epoch AI Benchmarking Hub | 12.7%† | 9.5% |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | – | 21.1% |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | – | 74.1% |
| Global-MMLU-Lite (AA) | Artificial Analysis | 90.6%† | – |
| Instruction following | |||
| LiveBench Language | LiveBench | 83.3% | 76.2%† |
| LiveBench Instruction Following | LiveBench | 63.3% | 62.3%† |
| IFBench | Artificial Analysis | 44.6%† | 73.3% |
| MultiChallenge | Scale AI SEAL | 56.0%† | – |
| TutorBench | Scale AI SEAL | 53.5%† | – |
| Arena Instruction Following | Arena (LMArena) | 1511 | 1470 |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1503 | 1475 |
| EQ-Bench 4 | EQ-Bench | 1210† | 1208† |
| EQ-Bench Creative Writing v3 | EQ-Bench | 1810† | 1759† |
| EQ-Bench Longform Writing | EQ-Bench | 77.7† | 77.9† |
| Arena Creative Writing | Arena (LMArena) | 1498 | 1451 |
| Arena Multi-turn | Arena (LMArena) | 1516 | 1477 |
| Arena Longer Queries | Arena (LMArena) | 1521 | 1486 |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1315 | – |
| MMMU-Pro | Artificial Analysis | 72.5%† | – |
| VISTA | Scale AI SEAL | 45.5%† | – |
| MMMU-Pro (Vals) | Vals AI | 83.9%† | – |
| MortgageTax | Vals AI | 68.5%† | – |
| VTB | Scale AI SEAL | 27.5%† | – |
| Furniture Assembly | Epoch AI Benchmarking Hub | 28.3%† | – |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 67.0%† | 78.3% |
| Arena Document | Arena (LMArena) | 1497 | – |
| MLCR | Artificial Analysis | – | 12.8% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 156.3† | 153.1† |
| LiveBench 1.0 | LiveBench | 74.5% | 73.2%† |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 26.4† | 33.7 |
| AA Progress Index v1 | Artificial Analysis | 274.5† | 302.5 |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | 20.5%† | 48.7%† |
| MASK | Scale AI SEAL | 96.3%† | – |
| Metadata | |||
| AA Openness Index | Artificial Analysis | – | 44.4 |
Data as of 2026-11-22. Best configuration of each model; every score links to its source on the model pages.
DuelLab GameBench 2, release gb2-20261119-1d7fd529faf6 (https://benchmarks.duellab.org/releases/gb2-20261119-1d7fd529faf6/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
What changed
- GLM 5.2 (max): 2 score revisions.
- GLM 5.2 (max): 1 new result (GDPval-AA v2.1).
- Claude Opus 4.6 (high): BenchLeader Index 63.0 → 62.5.
- GLM 5.2 (max): BenchLeader Index 59.5 → 59.0.
- GLM 5.2 (max): 3 new results (AIME 2026, HMMT February 2026, MathArena Apex); 1 score revision.
- GLM 5.2 (max): 1 new result (MirrorCode).
- GLM 5.2 (max): BenchLeader Index 60.2 → 59.5.
- GLM 5.2 (max): 4 new results (HLE-Diamond (AA), AA-Omniscience v1.1, AA-Omniscience v1.1: accuracy…).
- GLM 5.2 (max): cached input price moved from $0.30/M to $0.26/M (13% down).
- GLM 5.2 (max): 1 score revision; cached input price moved from $0.26/M to $0.30/M (15% up).
Related comparisons
- Claude Opus 4.6 vs Claude Fable 5.2
- Claude Opus 4.6 vs Gemini 4 Carbon
- Claude Opus 4.6 vs GPT-6.2 Astra
- Claude Opus 4.6 vs Gemini 4 Neon
- Claude Opus 4.6 vs GPT-6 Astra
- Claude Opus 4.6 vs Claude Opus 5.5
- Claude Opus 4.6 vs Claude Fable 5.1
- Claude Opus 4.6 vs GPT-6.1 Sol
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
Claude Opus 4.6 vs GLM 5.2: questions
- Is Claude Opus 4.6 better than GLM 5.2?
- Claude Opus 4.6 (high) and GLM 5.2 (max) are neck and neck on quality (62.4 vs 58.8, within the ±4.9 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 22 Nov 2026, so check the category scores for your use.
- Claude Opus 4.6 vs GLM 5.2: which is cheaper, and will that change?
- GLM 5.2 is cheaper today: $2.15 against $10.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is Claude Opus 4.6 or GLM 5.2 better for coding?
- They are level in coding (62 vs 62 on the category index). On Arena WebDev, the top shared coding benchmark, it is 1546 for Claude Opus 4.6 against 1608 for GLM 5.2.
- Which is faster, Claude Opus 4.6 or GLM 5.2?
- GLM 5.2 streams ~1.5× faster (58 vs 37 output tokens per second, Artificial Analysis 7-day median).
- Which is better for long documents?
- Both accept 1,000,000 tokens of context.
More questions (3)
- Which has lower latency?
- Claude Opus 4.6 answers first: 1.4 s against 34.1 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 Claude Opus 4.6 vs GLM 5.2?
- Across the 14 benchmarks both have, Claude Opus 4.6 leads 13, GLM 5.2 leads 1. Every overall index that rates both puts Claude Opus 4.6 ahead.
- When was GLM 5.2 released?
- GLM 5.2 was released on 16 Jun 2026.