Claude Opus 5 vs Gemini 4 Carbon
Full profiles: Claude Opus 5 (Anthropic) and Gemini 4 Carbon (Google): every benchmark, price and rank.
Updated · Data from 11 sources
As of 29 Oct 2026, Claude Opus 5 and Gemini 4 Carbon are neck and neck on quality (65.5 vs 65.1 on the BenchLeader Index, inside the ±6.4 margin), and Gemini 4 Carbon is 1.3× cheaper at $8.00 vs $10.00 per 1M tokens blended. Gemini 4 Carbon leads in every category we measure both in. Gemini 4 Carbon streams ~1.5× faster.
Key facts (as of 29 Oct 2026)
- Claude Opus 5 scores 65.5 on the BenchLeader Index and Gemini 4 Carbon scores 65.1 (BenchLeader, 29 Oct 2026).
- Claude Opus 5 costs $10.00 per 1M tokens blended (list price, 29 Oct 2026).
- Gemini 4 Carbon costs $8.00 per 1M tokens blended (list price, 29 Oct 2026).
- Gemini 4 Carbon streams ~1.5× faster than Claude Opus 5 (Artificial Analysis, 7-day median, 29 Oct 2026).
- Of the 10 benchmarks both models have, Gemini 4 Carbon leads 8 (BenchLeader, 29 Oct 2026).
Claude Opus 5 benchmarks max Anthropic
Gemini 4 Carbon benchmarks high GoogleOpen 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 |
|---|---|---|---|
| Coding | Gemini 4 Carbon | 72 vs 70 category score; SciCode 64.2% vs 56.4% | Medium |
| Reasoning | Gemini 4 Carbon | 79 vs 69 category score; Humanity's Last Exam (AA) 60.4% vs 54.9% | High |
| Knowledge | Gemini 4 Carbon | 76 vs 65 category score | High |
| Long documents | Either | 65 vs 65 category score; AA-LCR v1.1 79.3% vs 79.3% | Either |
| Chat quality (human preference) | – | Only Claude Opus 5 has an Arena rating (68 category score) | Not comparable |
| Lowest cost | Gemini 4 Carbon | $8.00 vs $10.00 per 1M blended; agentic coding session $0.320 vs $0.400 | – |
| Fastest first answer | Claude Opus 5 | 48.4 s (Claude Opus 5 (max)) vs 103.1 s (Gemini 4 Carbon (high)) 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 5 (max) and Gemini 4 Carbon (high) are neck and neck on quality (65.5 vs 65.1, within the ±6.4 margin).
- Claude Opus 5 (max) is stronger in agents & tools, human preference, maths, multimodal.
- Gemini 4 Carbon (high) is stronger in coding, knowledge, reasoning.
- Gemini 4 Carbon (high) is 1.3× cheaper ($8.00 vs $10.00 per 1M blended).
- Gemini 4 Carbon (high) streams 1.7× faster (100 vs 59 tokens per second).
At a glance
| Spec | Claude Opus 5 | Gemini 4 Carbon |
|---|---|---|
| Developer | Anthropic | |
| Released | 24 Jul 2026 | 28 Oct 2026 |
| Context window (tokens) | 1,000,000 | 1,000,000 |
| Compared at | Claude Opus 5 (max) | Gemini 4 Carbon (high) |
| Open weights | No | No |
| Price in / out / cached input per 1M | $5.00 / $25.00 / $0.500 | $4.00 / $20.00 / – |
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 5 (max) | Gemini 4 Carbon (high) | Leader |
|---|---|---|---|
| Reasoning | 69.2 | 78.9 | Gemini 4 Carbon |
| Coding | 70.0 | 71.7 | Gemini 4 Carbon |
| Agents & tools | 64.3 | – | not comparable |
| Maths | 70.4 | – | not comparable |
| Knowledge | 65.0 | 75.8 | Gemini 4 Carbon |
| Human preference | 68.4 | – | not comparable |
| Multimodal | 67.6 | – | not comparable |
| Long context | 64.5 | 64.5 | level |
Where sources disagree
Across the 10 benchmarks both have, Claude Opus 5 leads 1, Gemini 4 Carbon leads 8 and 1 is a tie.
| Overall index | Claude Opus 5 (max) | Gemini 4 Carbon (high) | Leader |
|---|---|---|---|
| BenchLeader Index | 65.5 | 65.1 | 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 | Claude Opus 5 (max) | Gemini 4 Carbon (high) |
|---|---|---|
| Blended price per 1M tokens | $10.00 | $8.00 |
| Input / output per 1M | $5.00 / $25.00 | $4.00 / $20.00 |
| Cached input per 1M | $0.500 | – |
| Price per task: chat reply | $0.0095 | $0.0076 |
| Price per task: summarize a 30-page report | $0.075 | $0.060 |
| Price per task: code edit | $0.068 | $0.054 |
| Price per task: agentic coding session | $0.400 | $0.320 |
| Price per task: structured extraction | $0.015 | $0.012 |
| Cost to run a full benchmark suite | – | $7.76 |
| Output speed | 59 tok/s | 100 tok/s |
| Time to first answer | 48.4 s | 103.1 s |
| Time to a complete answer | 56.9 s | 108.1 s |
Each model is compared at its best-ranked reasoning-effort setting. Claude Opus 5 (max) scores 65.5 and takes 48 s to its first answer token; Claude Opus 5 scores 65.2 in 3 s, Claude Opus 5 (high) scores 64.4 in 14 s. Gemini 4 Carbon (high) scores 65.1 and takes 103 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
131 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 5 (max) | Gemini 4 Carbon (high) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 65.5 | 65.1 |
| Agents & tools score | BenchLeader | 64.3 | – |
| Coding score | BenchLeader | 70.0 | 71.7 |
| Human preference score | BenchLeader | 68.4 | – |
| Knowledge score | BenchLeader | 65.0 | 75.8 |
| Long context score | BenchLeader | 64.5 | 64.5 |
| Maths score | BenchLeader | 70.4 | – |
| Multimodal score | BenchLeader | 67.6 | – |
| Reasoning score | BenchLeader | 69.2 | 78.9 |
| Blended price $/M | BenchLeader | $10.00 | $8.00 |
| Output speed | BenchLeader | 59 tok/s | 100 tok/s |
| Time to first answer | BenchLeader | 48.4 s | 103.1 s |
| Context window | BenchLeader | 1,000,000 tokens | 1,000,000 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 93.9% | – |
| SimpleBench | SimpleBench | 80.6%† | – |
| Arena Hard Prompts | Arena (LMArena) | 1515 | – |
| LiveBench 1.0 Reasoning | LiveBench | 91.2% | – |
| GPQA Diamond (AA) | Artificial Analysis | 93.2% | – |
| Humanity's Last Exam (AA) | Artificial Analysis | 54.9% | 60.4% |
| HLE-Diamond (AA) | Artificial Analysis | – | 69.1% |
| GPQA Diamond (Vals) | Vals AI | 93.4%† | – |
| ARC-AGI-1 | ARC Prize | 97.5% | – |
| ARC-AGI-2 | ARC Prize | 90.4% | – |
| ARC-AGI-3 | ARC Prize | 30.2%† | – |
| CritPt | Artificial Analysis | 29.1% | 36.0% |
| CritPt v2 hard subset (AA) | Artificial Analysis | – | 63.0% |
| MysteryMechanism | Vals AI | 37.4%† | – |
| Chess Puzzles | Epoch AI Benchmarking Hub | 43.0% | – |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 59.0% | – |
| LMCA | Epoch AI Benchmarking Hub | 63.3% | – |
| DTBench | Epoch AI Benchmarking Hub | 97.6% | – |
| ForecastBench | ForecastBench | 66.5% | – |
| BTF-3 | Epoch AI Benchmarking Hub | 12.0%† | – |
| Coding | |||
| SciCode | SciCode | 56.4% | 64.2% |
| WeirdML | WeirdML | 91.8% | – |
| FrontierCode | Cognition | 53.4% | – |
| Arena Coding | Arena (LMArena) | 1528 | – |
| Arena WebDev | Arena (LMArena) | 1690 | – |
Show all 131 benchmarks
| Metric | Publisher | Claude Opus 5 (max) | Gemini 4 Carbon (high) |
|---|---|---|---|
| Coding | |||
| LiveBench 1.0 Coding | LiveBench | 81.5% | – |
| LiveCodeBench | Vals AI | 89.0%† | – |
| IOI | Vals AI | 84.3%† | – |
| SWE-bench (Vals) | Vals AI | 97.0%† | – |
| Code Migration | Vals AI | 57.5%† | – |
| ProgramBench | Vals AI | 3.0%† | – |
| Vibe Code Bench 1-100 | Vals AI | 28.5%† | – |
| Vibe Code Bench v1.1 | Vals AI | 88.4%† | 89.1% |
| SWE Atlas: Codebase QnA | Scale AI SEAL | 63.2%† | – |
| SWE Atlas: Test Writing | Scale AI SEAL | 62.2%† | – |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | 73.7% | – |
| CursorBench | Cursor | 46.6% | – |
| ALE-Bench | Epoch AI Benchmarking Hub | 2164.6† | – |
| FrontierSWE | Epoch AI Benchmarking Hub | 52.0% | – |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 49.0% | 65.2% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 89.1% | – |
| Startrise Frontend | Startrise | 82.3%† | – |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 48.5% |
| FrontierCode 2.0 (AA) | Artificial Analysis | – | 40.6% |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 51.8% | – |
| OSWorld-Verified 2.0 | OSWorld | 31.4% | – |
| APEX-Agents | Mercor | 65.8% | – |
| Arena Agent | Arena (LMArena) | 8.2 | – |
| LiveBench Agentic Coding | LiveBench | 65.2% | – |
| Terminal-Bench 2.1 (Vals) | Vals AI | 84.6%† | – |
| MCP Atlas | Scale AI SEAL | 85.8%† | – |
| HiL-Bench | Scale AI SEAL | 57.0%† | – |
| τ³-bench Banking | τ²-bench | 48.7% | – |
| τ³-Banking (AA) | Artificial Analysis | 42.1% | – |
| Analyst Agent (AA) | Artificial Analysis | 53.8% | – |
| CUA-bench | Vals AI | 9.0% | – |
| CyberBench | Vals AI | 65.4%† | – |
| Finance Agent v2 | Vals AI | 58.6%† | – |
| Finance Agent v2.1 | Vals AI | – | 67.2% |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | – | 17.5% |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 6.7%† | – |
| Legal Research Bench | Vals AI | 55.3%† | 55.8% |
| SkillsBench | Vals AI | 60.4%† | – |
| SREBench | Vals AI | 12.2%† | – |
| Tax Agent Bench | Vals AI | 45.5%† | 48.9% |
| Terminal-Bench 4.0 (Vals) | Vals AI | 53.5%† | 63.1% |
| Terminal-Bench Science | Vals AI | 27.1%† | – |
| Time Horizon Index: KSP | Vals AI | 18.8%† | – |
| Arena Search | Arena (LMArena) | 1214† | – |
| BALROG | BALROG | 63.4% | – |
| PostTrainBench | Epoch AI Benchmarking Hub | 35.0%† | – |
| GBAEval | Epoch AI Benchmarking Hub | 79.6%† | – |
| Vending-Bench 2 | Andon Labs | 11181.9† | 16484.2† |
| AutomationBench | Artificial Analysis | 56.6% | 77.8% |
| GDP.pdf (AA) | Artificial Analysis | 21.6% | 28.4% |
| EnterpriseOps-Gym | Artificial Analysis | 47.5% | – |
| AA-Briefcase v1.1 | Artificial Analysis | 1655 | 1609 |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 28.6% | 64.8% |
| DeepSec-Bench (AA) | Artificial Analysis | 25.9% | – |
| FACTS Search | 85.3%† | – | |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 85.6% | – |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 73.2% | – |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 98.9% | – |
| ProofBench | Vals AI | 99.0% | – |
| LiveBench Mathematics | LiveBench | 95.7% | – |
| Arena Maths | Arena (LMArena) | 1517 | – |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 59.9% | – |
| LiveBench 1.0 Data Analysis | LiveBench | 74.5% | – |
| AA-Omniscience v1.1 | Artificial Analysis | – | 53.5 |
| MMLU-Pro | Vals AI | 91.6%† | – |
| LegalBench | Vals AI | 87.0%† | – |
| CorpFin | Vals AI | 73.2%† | – |
| TaxEval | Vals AI | 75.1%† | – |
| BioMysteryBench | Vals AI | 79.3%† | – |
| Excel Modeling Benchmark | Vals AI | 73.6%† | – |
| Excel Modeling Benchmark v1.1 | Vals AI | – | 76.5% |
| MedCode | Vals AI | 63.6%† | – |
| MedScribe | Vals AI | 91.0%† | – |
| Public Benefits Bench v1.1 | Vals AI | 76.9%† | – |
| DrugDiscoveryBench | Scale AI SEAL | 53.7% | – |
| EBR-bench | Epoch AI Benchmarking Hub | 45.7% | – |
| GDP.pdf | Epoch AI Benchmarking Hub | 24.0% | – |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | – | 58.3% |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | – | 88.6% |
| Instruction following | |||
| LiveBench Language | LiveBench | 88.7% | – |
| LiveBench Instruction Following | LiveBench | 63.8% | – |
| Arena Instruction Following | Arena (LMArena) | 1495 | – |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1491 | – |
| EQ-Bench 4 | EQ-Bench | 1365† | – |
| Arena Creative Writing | Arena (LMArena) | 1474 | – |
| Arena Multi-turn | Arena (LMArena) | 1487 | – |
| Arena Longer Queries | Arena (LMArena) | 1503 | – |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1320† | – |
| MMMU-Pro | Artificial Analysis | 84.7% | – |
| MMMU-Pro (Vals) | Vals AI | 89.9%† | – |
| MortgageTax | Vals AI | 72.1%† | – |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | 30.4%† | 58.6%† |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 79.3% | 79.3% |
| Arena Document | Arena (LMArena) | 1490† | – |
| MLCR | Artificial Analysis | 11.1% | – |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 163.9† | – |
| LiveBench 1.0 | LiveBench | 80.1% | – |
| AA Intelligence Index v5.0 | Artificial Analysis | – | 59.4 |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 50.8 | – |
| AA Progress Index v1 | Artificial Analysis | 333.7 | 351.6 |
| Vals Index | Vals AI | – | 70.4 |
| Vals Finance Index | Vals AI | – | 66.3% |
| Vals RSI Index | Vals AI | 33.0% | – |
| Vals Science Index | Vals AI | 47.9%† | – |
| Vals Multimodal Index | Vals AI | 73.9%† | – |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | 16.8%† | – |
Data as of 2026-10-29. Best configuration of each model; every score links to its source on the model pages.
What changed
- Claude Opus 5 (max): BenchLeader Index 66.3 → 65.5.
- Claude Opus 5 (max): 1 new result (AA Progress Index v1).
- Claude Opus 5 (max): BenchLeader Index 66.8 → 66.3.
- Claude Opus 5 (max): 1 new result (Vals RSI Index).
- Claude Opus 5 (max): 1 score revision.
- Claude Opus 5 (max): 1 new result (ForecastBench); 1 score revision.
- Claude Opus 5 (max): 4 new results (GDPval-AA v2.1, AA-Omniscience v1.1: accuracy, AA-Omniscience v1.1: non-hallucination…).
- Claude Opus 5 (max): 1 score revision.
- Claude Opus 5 (max): 2 score revisions.
- Claude Opus 5 (max): 2 new results (Terminal-Bench-Science 0.1 (AA), DeepSec-Bench (AA)).
Related comparisons
- Claude Opus 5 vs Claude Opus 5.5
- Claude Opus 5 vs GPT-6 Astra
- Claude Opus 5 vs GPT-6.1 Sol
- Gemini 4 Carbon vs Claude Opus 5.5
- Gemini 4 Carbon vs GPT-6 Astra
- Gemini 4 Carbon vs GPT-6.1 Sol
- Gemini 4 Carbon vs Claude Fable 5.1
- Claude Sonnet 5 vs Gemini 4 Carbon
Rankings: Best LLM overall · Best LLM for coding · Best LLM for agents · Best cheap LLMs
Claude Opus 5 vs Gemini 4 Carbon: questions
- Is Claude Opus 5 better than Gemini 4 Carbon?
- Claude Opus 5 (max) and Gemini 4 Carbon (high) are neck and neck on quality (65.5 vs 65.1, within the ±6.4 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.
- Claude Opus 5 vs Gemini 4 Carbon: which is cheaper, and will that change?
- Gemini 4 Carbon is cheaper today: $8.00 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 5 or Gemini 4 Carbon better for coding?
- Gemini 4 Carbon scores higher in coding (72 vs 70 on the category index, where 50 is average). On SciCode, the top shared coding benchmark, it is 56.4% for Claude Opus 5 against 64.2% for Gemini 4 Carbon.
- Which is faster, Claude Opus 5 or Gemini 4 Carbon?
- Gemini 4 Carbon streams ~1.5× faster (100 vs 59 output tokens per second, Artificial Analysis 7-day median).
- Which is better for long documents?
- They are level on long-context benchmarks (65 vs 65). Both accept 1,000,000 tokens of context.
More questions (3)
- Which has lower latency?
- Claude Opus 5 answers first: 48.4 s against 103.1 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 5 vs Gemini 4 Carbon?
- Across the 10 benchmarks both have, Claude Opus 5 leads 1, Gemini 4 Carbon leads 8 and 1 is a tie.
- When was Gemini 4 Carbon released?
- Gemini 4 Carbon was released on 28 Oct 2026.