GPT-6 Astra vs Claude Opus 5
Wins 5 of 8 areas
Reasoning · Facts · Math · Long documents · Following instructions
Wins 1 of 8 areas
Coding
GPT-6 Astra is the stronger all-rounder.Claude Opus 5 is cheaper and better at coding.
Both rank among the ten best models we track in reasoning and math.
Scores updated · 44 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking60GPT-6 Astra6 of 7 tests · 1 tie
- FactsGetting facts right instead of making them up30GPT-6 Astra3 of 3 tests
- MathCompetition and research-level math30GPT-6 Astra3 of 3 tests
- Long documentsFinding answers in very long texts10GPT-6 Astra1 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-6 Astra1 of 1 test
- CodingWriting and fixing software12Claude Opus 52 of 5 tests · 2 ties
- AgentsCarrying out multi-step tasks on its own11Even1 each · 2 ties
- Images and chartsUnderstanding pictures, charts and video11Even1 each
Long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
Claude Opus 5 costs 50% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-6 Astra pulls ahead
- Solves new puzzle games it has never seenARC-AGI-3+69.7points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+24.4points ahead
- Short factual questions, answered correctlySimpleQA Verified+15.7points ahead
Where Claude Opus 5 pulls ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+7.9points ahead
- Real work tasks from 44 professionsGDPVal+7.5points ahead
- Programming problems, refreshed regularlyLiveBench · Coding+1.1points ahead
Every test, side by side
All 44 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 5
- LMArena · WebDevGPT-6 Astra by 86 rating points17761690+86 rating
- LiveBench · Agentic CodingClaude Opus 5 by 7.957.365.2+7.9
- LiveBench · CodingClaude Opus 5 by 1.180.481.5+1.1
- Terminal-Bench 2.1tie88.489.1tie
- SciCodetie56.556.4tie
AgentsEven
- Terminal-Bench 4.0GPT-6 Astra by 10.159.149+10.1
- GDPValClaude Opus 5 by 7.553.661.1+7.5
- BrowseComptie91.590.8tie
- τ-Bench V3 · Bankingtie41.442.1tie
ReasoningGPT-6 Astra
- ARC-AGI-3GPT-6 Astra by 69.799.930.2+69.7
- ARC-AGI-2GPT-6 Astra by 4.69590.4+4.6
- SimpleBenchGPT-6 Astra by 383.680.6+3
- GPQA DiamondGPT-6 Astra by 2.996.193.2+2.9
- CritPtGPT-6 Astra by 2.631.729.1+2.6
- LiveBench · ReasoningGPT-6 Astra by 1.592.791.2+1.5
- Humanity's Last Examtie54.754.9tie
FactsGPT-6 Astra
- SimpleQA VerifiedGPT-6 Astra by 15.775.659.9+15.7
- AA-Omniscience · Non-hallucinationGPT-6 Astra by 9.548.739.2+9.5
- AA-Omniscience · AccuracyGPT-6 Astra by 1.762.660.9+1.7
Images and chartsEven
- LMArena · VisionClaude Opus 5 by 38 rating points12811319+38 rating
- MMMU-ProGPT-6 Astra by 2.286.984.7+2.2
MathGPT-6 Astra
- FrontierMath Tier 4GPT-6 Astra by 24.497.673.2+24.4
- FrontierMath Tiers 1-3 (v2)GPT-6 Astra by 8.193.785.6+8.1
- LiveBench · MathematicsGPT-6 Astra by 1.196.895.7+1.1
Following instructionsGPT-6 Astra
- LiveBench · Instruction FollowingGPT-6 Astra by 11.875.663.8+11.8
Other results18 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- Terminal-Bench-Science 0.1GPT-6 Astra by 35.664.629+35.6
- Agents' Last ExamGPT-6 Astra by 30.759.328.6+30.7
- GDPval-AA 2.1Claude Opus 5 by 166 rating points15421708+166 rating
- LVBenchGPT-6 Astra by 12.187.575.4+12.1
- livebench_data_analysisGPT-6 Astra by 8.48374.5+8.4
- HiL-Bench (Tools-allowed)GPT-6 Astra by 8.365.357+8.3
- AA-OmniscienceGPT-6 Astra by 6.343.437.1+6.3
- Terminal-Bench 4.0GPT-6 Astra by 5.957.751.8+5.9
- DeepSWE 1.1GPT-6 Astra by 5.374.168.8+5.3
- HealthBench Professional length-adjustedGPT-6 Astra by 4.964.759.8+4.9
- AA Agentic IndexClaude Opus 5 by 4.751.556.2+4.7
- HealthBench ProfessionalGPT-6 Astra by 3.663.459.8+3.6
- OSWorld 2.0GPT-6 Astra by 272.670.6+2
- AA IntelligenceGPT-6 Astra by 1.952.750.8+1.9
- Artificial Analysis Coding IndexClaude Opus 5 by 1.176.978+1.1
- livebench_languagetie89.488.7tie
- HealthBench length-adjustedtie58.357.8tie
- ARC-AGI-1tie97.597.5tie
Questions people ask
Which is better, GPT-6 Astra or Claude Opus 5?
GPT-6 Astra wins five of the eight areas we test: reasoning, facts, math, long documents and following instructions. Claude Opus 5 wins coding, and costs 50% less. They are level on agents and images and charts.
Which is better for coding?
Claude Opus 5. It wins 2 of the 5 coding tests both models report; GPT-6 Astra wins 1, and 2 are ties.
Which is cheaper?
GPT-6 Astra costs $10.00 per million input tokens and $50.00 per million output tokens; Claude Opus 5 costs $5.00 and $25.00. That makes Claude Opus 5 about 50% cheaper for the same work.
How do you compare the two?
We use the 44 benchmark tests both models have published scores on. The verdict counts the 26 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 18 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.