Gemini 3.8 Flash vs GPT-6 Astra
Wins 1 of 8 areas
Following instructions
Wins 4 of 8 areas
Coding · Reasoning · Facts · Math
GPT-6 Astra wins more areas, narrowly.Gemini 3.8 Flash is cheaper and better at following instructions.
Both rank among the ten best models we track in reasoning and facts.
Scores updated · 38 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 thinking06GPT-6 Astra6 of 7 tests · 1 tie
- CodingWriting and fixing software03GPT-6 Astra3 of 5 tests · 2 ties
- FactsGetting facts right instead of making them up03GPT-6 Astra3 of 3 tests
- MathCompetition and research-level math03GPT-6 Astra3 of 3 tests
- Following instructionsDoing exactly what it is asked10Gemini 3.8 Flash1 of 1 test
- AgentsCarrying out multi-step tasks on its own22Even2 each
- Images and chartsUnderstanding pictures, charts and video11Even1 each
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
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.
Gemini 3.8 Flash costs 93% 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+89.5points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+75.6points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+39.4points ahead
Where Gemini 3.8 Flash pulls ahead
- Follows detailed instructions exactlyLiveBench · Instruction Following+5.8points ahead
- Finds the root cause of IT system incidentsAA IT-Bench SRE+3.9points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+3.5points ahead
Every test, side by side
All 38 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 190 rating points15861776+190 rating
- LiveBench · CodingGPT-6 Astra by 7.972.580.4+7.9
- LiveBench · Agentic CodingGPT-6 Astra by 3.154.257.3+3.1
- Terminal-Bench 2.1tie87.688.4tie
- SciCodetie56.656.5tie
AgentsEven
- Terminal-Bench 4.0GPT-6 Astra by 39.419.759.1+39.4
- GDPValGPT-6 Astra by 6.846.853.6+6.8
- AA IT-Bench SREGemini 3.8 Flash by 3.952.548.6+3.9
- τ-Bench V3 · BankingGemini 3.8 Flash by 3.544.941.4+3.5
ReasoningGPT-6 Astra
- ARC-AGI-3GPT-6 Astra by 89.510.499.9+89.5
- CritPtGPT-6 Astra by 13.418.331.7+13.4
- Humanity's Last ExamGPT-6 Astra by 6.947.854.7+6.9
- ARC-AGI-2GPT-6 Astra by 5.889.295+5.8
- LiveBench · ReasoningGPT-6 Astra by 3.489.392.7+3.4
- SimpleBenchGPT-6 Astra by 1.282.483.6+1.2
- GPQA Diamondtie95.396.1tie
FactsGPT-6 Astra
- AA-Omniscience · AccuracyGPT-6 Astra by 854.662.6+8
- SimpleQA VerifiedGPT-6 Astra by 5.969.775.6+5.9
- AA-Omniscience · Non-hallucinationGPT-6 Astra by 3.944.848.7+3.9
Images and chartsEven
- LMArena · VisionGemini 3.8 Flash by 33 rating points13141281+33 rating
- MMMU-ProGPT-6 Astra by 1.385.686.9+1.3
MathGPT-6 Astra
- FrontierMath Tier 4GPT-6 Astra by 75.621.997.6+75.6
- FrontierMath Tiers 1-3 (v2)GPT-6 Astra by 25.368.493.7+25.3
- LiveBench · MathematicsGPT-6 Astra by 5.291.696.8+5.2
Following instructionsGemini 3.8 Flash
- LiveBench · Instruction FollowingGemini 3.8 Flash by 5.881.475.6+5.8
Other results12 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 4.0GPT-6 Astra by 38.619.157.7+38.6
- livebench_data_analysisGPT-6 Astra by 295483+29
- HiL-Bench (Tools-allowed)GPT-6 Astra by 23.841.565.3+23.8
- AA-OmniscienceGPT-6 Astra by 13.829.643.4+13.8
- OSWorld 2.0GPT-6 Astra by 13.65972.6+13.6
- AA IntelligenceGPT-6 Astra by 11.840.952.7+11.8
- AA Agentic IndexGPT-6 Astra by 10.441.151.5+10.4
- livebench_languageGPT-6 Astra by 1.687.889.4+1.6
- ARC-AGI-1Gemini 3.8 Flash by 198.597.5+1
- Artificial Analysis Coding Indextie76.376.9tie
- DeepSWE 1.1tie73.774.1tie
- LVBenchtie87.187.5tie
Questions people ask
Which is better, Gemini 3.8 Flash or GPT-6 Astra?
GPT-6 Astra wins four of the eight areas we test: coding, reasoning, facts and math. Gemini 3.8 Flash wins following instructions, and costs 93% less. They are level on agents, images and charts and long documents.
Which is better for coding?
GPT-6 Astra. It wins 3 of the 5 coding tests both models report; Gemini 3.8 Flash wins none, and 2 are ties.
Which is cheaper?
Gemini 3.8 Flash costs $0.75 per million input tokens and $3.75 per million output tokens; GPT-6 Astra costs $10.00 and $50.00. That makes Gemini 3.8 Flash about 93% cheaper for the same work.
How do you compare the two?
We use the 38 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 12 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.