DeepSeek-V4-Pro vs GPT-6 Astra
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Math · Long documents · Following instructions
GPT-6 Astra is the stronger all-rounder.DeepSeek-V4-Pro is cheaper.
Scores updated · 37 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software05GPT-6 Astra5 of 5 tests
- AgentsCarrying out multi-step tasks on its own05GPT-6 Astra5 of 5 tests
- ReasoningHard problems that need careful thinking05GPT-6 Astra5 of 5 tests
- FactsGetting facts right instead of making them up03GPT-6 Astra3 of 4 tests · 1 tie
- MathCompetition and research-level math03GPT-6 Astra3 of 3 tests
- Long documentsFinding answers in very long texts01GPT-6 Astra1 of 1 test
- Following instructionsDoing exactly what it is asked01GPT-6 Astra1 of 1 test
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.
DeepSeek-V4-Pro costs 98% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where GPT-6 Astra pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+95.2points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+48.4points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+44.5points ahead
Where DeepSeek-V4-Pro pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 37 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-6 Astra
- LMArena · WebDevGPT-6 Astra by 311 rating points14641775+311 rating
- Terminal-Bench 2.1GPT-6 Astra by 24.46488.4+24.4
- LiveBench · Agentic CodingGPT-6 Astra by 14.742.657.3+14.7
- LiveBench · CodingGPT-6 Astra by 10.47080.4+10.4
- SciCodeGPT-6 Astra by 5.750.856.5+5.7
AgentsGPT-6 Astra
- Terminal-Bench 4.0GPT-6 Astra by 44.514.659.1+44.5
- GDPValGPT-6 Astra by 20.832.853.6+20.8
- τ-Bench V3 · BankingGPT-6 Astra by 11.330.141.4+11.3
- AA IT-Bench SREGPT-6 Astra by 10.338.348.6+10.3
- BrowseCompGPT-6 Astra by 8.183.491.5+8.1
ReasoningGPT-6 Astra
- SimpleBenchGPT-6 Astra by 32.750.983.6+32.7
- CritPtGPT-6 Astra by 18.812.931.7+18.8
- Humanity's Last ExamGPT-6 Astra by 17.237.554.7+17.2
- LiveBench · ReasoningGPT-6 Astra by 1082.792.7+10
- GPQA DiamondGPT-6 Astra by 7.388.896.1+7.3
FactsGPT-6 Astra
- AA-Omniscience · Non-hallucinationGPT-6 Astra by 42.85.948.7+42.8
- SimpleQA VerifiedGPT-6 Astra by 29.446.275.6+29.4
- AA-Omniscience · AccuracyGPT-6 Astra by 19.64362.6+19.6
- Vectara HHEM hallucination ratelower is bettertie8.68.7tie
MathGPT-6 Astra
- FrontierMath Tier 4GPT-6 Astra by 95.22.497.6+95.2
- FrontierMath Tiers 1-3 (v2)GPT-6 Astra by 48.445.393.7+48.4
- LiveBench · MathematicsGPT-6 Astra by 6.190.796.8+6.1
Following instructionsGPT-6 Astra
- LiveBench · Instruction FollowingGPT-6 Astra by 13.262.475.6+13.2
Other results13 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.
- AA-OmniscienceGPT-6 Astra by 54.1-10.743.4+54.1
- Terminal-Bench 4.0GPT-6 Astra by 45.312.457.7+45.3
- Agents' Last ExamGPT-6 Astra by 33.625.759.3+33.6
- HiL-Bench (Tools-allowed)GPT-6 Astra by 26.838.565.3+26.8
- AA Agentic IndexGPT-6 Astra by 23.827.751.5+23.8
- AA IntelligenceGPT-6 Astra by 22.330.452.7+22.3
- Artificial Analysis Coding IndexGPT-6 Astra by 17.559.476.9+17.5
- livebench_languageGPT-6 Astra by 11.378.189.4+11.3
- livebench_data_analysisGPT-6 Astra by 8.574.583+8.5
- HLE (with tools)DeepSeek-V4-Pro by 2.86057.2+2.8
- vectara_answer_rateGPT-6 Astra by 2.897.2100+2.8
- vectara_avg_summary_lengthtie153.8148.5tie
- vectara_factual_consistencytie91.491.3tie
Questions people ask
Which is better, DeepSeek-V4-Pro or GPT-6 Astra?
GPT-6 Astra wins all seven areas where both have results: coding, agents, reasoning, facts, math, long documents and following instructions. DeepSeek-V4-Pro wins none, but costs 98% less.
Which is better for coding?
GPT-6 Astra. It wins 5 of the 5 coding tests both models report; DeepSeek-V4-Pro wins none.
Which is cheaper?
DeepSeek-V4-Pro costs $0.43 per million input tokens and $0.87 per million output tokens; GPT-6 Astra costs $10.00 and $50.00. That makes DeepSeek-V4-Pro about 98% cheaper for the same work.
How do you compare the two?
We use the 37 benchmark tests both models have published scores on. The verdict counts the 24 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 13 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.