Gemma 4 26B A4B vs GPT-5
Wins 0 of 8 areas
—
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents
GPT-5 is the stronger all-rounder.Gemma 4 26B A4B is cheaper.
Scores updated · 41 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.
- CodingWriting and fixing software17GPT-57 of 8 tests
- AgentsCarrying out multi-step tasks on its own03GPT-53 of 3 tests
- ReasoningHard problems that need careful thinking03GPT-53 of 3 tests
- Images and chartsUnderstanding pictures, charts and video13GPT-53 of 4 tests
- FactsGetting facts right instead of making them up12GPT-52 of 3 tests
- MathCompetition and research-level math01GPT-51 of 1 test
- Long documentsFinding answers in very long texts01GPT-51 of 1 test
- Following instructionsDoing exactly what it is asked00Even0 each · 1 tie
Math, 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.
Gemma 4 26B A4B costs 95% 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-5 pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+57.5points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+38points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+28.6points ahead
Where Gemma 4 26B A4B pulls ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+3.8points ahead
Every test, side by side
All 41 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5
- SWE-bench VerifiedGPT-5 by 57.517.474.9+57.5
- SWE-bench MultilingualGPT-5 by 3817.355.3+38
- SWE-bench ProGPT-5 by 2813.841.8+28
- Terminal-Bench HardGPT-5 by 1913.632.6+19
- LiveCodeBench v6GPT-5 by 9.977.187+9.9
- LMArena · WebDevGPT-5 by 59 rating points13591418+59 rating
- Terminal-Bench 2.1Gemma 4 26B A4B by 3.83935.2+3.8
- SciCodeGPT-5 by 2.94042.9+2.9
AgentsGPT-5
- BrowseCompGPT-5 by 28.626.354.9+28.6
- GDPValGPT-5 by 17.63.421+17.6
- τ-Bench V3 · BankingGPT-5 by 10.11222.1+10.1
ReasoningGPT-5
- Humanity's Last ExamGPT-5 by 9.219.328.5+9.2
- GPQA DiamondGPT-5 by 6.279.285.4+6.2
- CritPtGPT-5 by 5.705.7+5.7
FactsGPT-5
- AA-Omniscience · AccuracyGPT-5 by 21.219.140.3+21.2
- Vectara HHEM hallucination ratelower is betterGemma 4 26B A4B by 9.95.215.1+9.9
- AA-Omniscience · Non-hallucinationGPT-5 by 4.213.617.8+4.2
Images and chartsGPT-5
- CharXiv (RQ)GPT-5 by 12.16981.1+12.1
- MMMUGPT-5 by 5.878.484.2+5.8
- LMArena · VisionGemma 4 26B A4B by 52 rating points12601208+52 rating
- MMMU-ProGPT-5 by 569.274.2+5
Other results17 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.
- HumanEvalGPT-5 by 43.45093.4+43.4
- AA-OmniscienceGPT-5 by 42.1-50.8-8.7+42.1
- τ²-Bench Telecom (AA run)GPT-5 by 41.243.684.8+41.2
- HLE (with tools)GPT-5 by 1817.235.2+18
- AA Agentic IndexGPT-5 by 15.51126.5+15.5
- τ²-BenchGPT-5 by 14.268.282.4+14.2
- vectara_factual_consistencyGemma 4 26B A4B by 9.994.884.9+9.9
- vectara_avg_summary_lengthGPT-5 by 95.6 rating points67.1162.7+95.6 rating
- AA IntelligenceGPT-5 by 6.316.723+6.3
- MMLU-ProGPT-5 by 4.582.687.1+4.5
- GPQA (unspecified)GPT-5 by 3.482.385.7+3.4
- HMMT Feb. 2025Gemma 4 26B A4B by 3.491.788.3+3.4
- MMLU-ReduxGPT-5 by 2.692.795.3+2.6
- HMMT Nov. 2025GPT-5 by 1.787.589.2+1.7
- Artificial Analysis Coding IndexGemma 4 26B A4B by 1.539.337.8+1.5
- Terminal-Bench 2.0GPT-5 by 134.235.2+1
- vectara_answer_ratetie99.899.9tie
Questions people ask
Which is better, Gemma 4 26B A4B or GPT-5?
GPT-5 wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and long documents. Gemma 4 26B A4B wins none, but costs 95% less. They are level on following instructions.
Which is better for coding?
GPT-5. It wins 7 of the 8 coding tests both models report; Gemma 4 26B A4B wins 1.
Which is cheaper?
Gemma 4 26B A4B costs $0.13 per million input tokens and $0.40 per million output tokens; GPT-5 costs $1.25 and $10.00. That makes Gemma 4 26B A4B about 95% cheaper for the same work.
How do you compare the two?
We use the 41 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 17 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.