Models / Gemma 4 26B A4B

Google

Gemma 4 26B A4B

Gemma 4 26B A4B is strongest at sticking to the facts; worth a look for simple, high-volume work where cost matters most. The cheapest model of similar intelligence is Qwen3.5-9B, at $0.11 per million tokens against $0.14.

Results 16 Nov 2026 · 37 results · 4 sources

Test Gemma 4 26B A4B on your prompt → It runs beside the models you could use instead, with every answer checked and costed.

Price per million tokens
$0.09 in · $0.30 out
Typical host on OpenRouter, 10 Nov 2026 · compare prices
Speed
82 tokens a second; on its test prompt the answer starts after 0.6 s, thinking time included
Artificial Analysis, on its usual API, at no reasoning; 53–82 across 2 settings
Intelligence Index
16.7
Artificial Analysis; 13.1–16.7 across 2 settings
Spring Prompt overall
Not ranked yet
needs our benchmarks and two groups of results
Developer
Google, based in the US
Weights published
Released
2 Apr 2026
262,144 tokens of context

Where it stands among the models you could choose

Available models grouped by price, intelligence, speed and developer location, plus models with published weights. Retired models are excluded.

AmongIntelligence IndexSpring Prompt overallPrice, blendedSpeed
Models available today 106th of 185Claude Opus 5.5 – $0.14 blended, 11th percentilecheapest: Mistral Nemo 43rd of 65Nemotron 3.5 Super VL
Models priced under $0.25 per million tokens 8th of 41Claude Haiku 5.5 – sets this group 13th of 19Nemotron 3.5 Super VL
Models with similar intelligence sets this group – $0.14 blended, 7th percentilecheapest: Qwen3.5-9B 7th of 8Nova 2 Lite
Models writing 50–100 tokens a second 14th of 21Claude Opus 5.5 – $0.14 blended, 10th percentilecheapest: Granite 4.2 8B sets this group
Models from developers based in the US 59th of 92Claude Opus 5.5 – $0.14 blended, 13th percentilecheapest: Gemma 3 4B 25th of 33Nemotron 3.5 Super VL
Open-weights models 45th of 104MiMo-V2.6-Pro – $0.14 blended, 16th percentilecheapest: Mistral Nemo 27th of 43Nemotron 3.5 Super VL

The name under each rank is the leader of that group; green is the top quarter of the group and red the bottom quarter, by rank. Price is shown as a percentile: the share of the group that costs less, so lower is cheaper. "Sets this group" marks the measure the group is defined by. Spring Prompt overall is our score out of 100 across every source (how it works). Price is the developer's list price (or the typical OpenRouter host where we haven't read one) for three input tokens to one output; intelligence and speed are from Artificial Analysis (data sourced from Artificial Analysis); "similar intelligence" means within 4 points on its Intelligence Index. Developer locations are where each company is based, not where a model is served. Groups under 3 models are not ranked.

How good is it, and for what?

Each use case ranks the available models its benchmarks measured, then those priced under $0.25 per million tokens. The bar shows its percentile in that field, best to the right. Open a row for the results behind it.

Product listingsTurning a sparse product feed and photos into listings that can go live Not measured
  • CatalogBench: has not measured this model.
Decks from an analysisTurning a finished analysis into a deck you could present as it is Not measured
  • DeckBench: has not measured this model.
User surveysPlanning a user survey and reading its results without being misled Not measured
  • SurveyBench: has not measured this model.
Marketing planningPlanning a year of ad spend without overspending Not measured
  • ROASBench: has not measured this model.
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 105th of 159 available · 11th of 37 at its price
BenchmarkGemma 4 26B A4BRank among availableBest available
Artificial AnalysisArtificial Analysis: Terminal-Bench 2.1 · reported 39.0% 71st of 107 Claude Fable 5.1 91.4%
Artificial AnalysisArtificial Analysis: τ-bench banking · reported 12.0% 66th of 104 Grok 4.6 50.7%
  • tau2-bench: has not measured this model.
  • Berkeley Function Calling Leaderboard (BFCL) V4: pending. Its latest results are dated 16 Dec 2025, before this model was released.
  • OpenHands Index: has not measured this model.
  • Microsoft STATE-Bench: has not measured this model.
  • Vending-Bench 2: has not measured this model.
Professional workReal tasks from banking, consulting and law, business documents and freelance projects Not measured
  • APEX-Agents: has not measured this model.
  • GDP.pdf: has not measured this model.
  • Remote Labor Index: has not measured this model.
Reasoning and knowledgeHard questions across science, maths and general knowledge 110th of 187 available · 10th of 41 at its price
BenchmarkGemma 4 26B A4BRank among availableBest available
Artificial AnalysisArtificial Analysis: Artificial Analysis Intelligence Index · reported 16.7 106th of 185 Claude Opus 5.5 57.6
Artificial AnalysisArtificial Analysis: Humanity's Last Exam · reported 19.3% 105th of 186 Claude Opus 5.5 61.4%
Artificial AnalysisArtificial Analysis: GPQA Diamond · reported 79.2% 96th of 168 GPT-6 Astra 96.3%
WritingWhat people prefer in blind comparisons, and judged writing quality 82nd of 169 available · 6th of 39 at its price
BenchmarkGemma 4 26B A4BRank among availableBest available
Arena (formerly LMArena)Arena Text: overall · reported 1,437 67th of 153 Gemini 4 Argon 1,525
UGI LeaderboardUGI: writing score · reported 43.8 67th of 124 Gemini 3.8 Flash 78.6
Sticking to the factsSummarising without inventing things, and factual answers 43rd of 139 available · 11th of 30 at its price
BenchmarkGemma 4 26B A4BRank among availableBest available
Vectara Hallucination LeaderboardVectara: hallucination rate · reported 5.2% 8th of 71 GPT-5.4 nano 3.1%
Arena (formerly LMArena)Arena Text factuality: overall · reported 1,445 59th of 109 Gemini 4 Argon 1,511
  • SimpleQA Verified (Epoch AI): has not measured this model.
Speed under pressureGood decisions against a real clock (fast chess) Not measured
  • BulletBench: has not measured this model.

Nearest alternatives

Against Gemma 4 26B A4B at $0.14 blended, Intelligence Index 16.7. "Similar intelligence" means within 4 points or better.

Where to run it

15 hosts on OpenRouter at the standard service tier, with a median of $0.14 blended per million tokens; quantisation where the host states it: BF16, FP4, FP8. Cheapest first.

HostQuantisationInputOutputBlended
Darkbloomnot stated$0.04$0.15$0.07
PrimeIntellectBF16$0.05$0.15$0.08
Makoranot stated$0.08$0.10$0.08
AkashMLFP4$0.07$0.24$0.11
DekaLLMBF16$0.06$0.33$0.13
Io NetBF16$0.08$0.28$0.13
DeepInfraFP8$0.07$0.34$0.14
NextBitBF16$0.09$0.30$0.14

Prices per million tokens from OpenRouter's public endpoint list, checked 10 Nov 2026; the 7 dearer hosts are left out.

Against the alternatives

The models you would most likely weigh it against: the leaders of the groups above.

Scroll sideways to see every alternative.

Gemma 4 26B A4BClaude Haiku 5.5
most intelligent at under $0.25 per million tokens
Qwen3.5-9B
cheapest with similar intelligence
Nova 2 Lite
fastest with similar intelligence
Claude Opus 5.5
most intelligent available today
Price per million tokens $0.14 $0.20$0.11$0.85$8.00
Tokens a second 82 24011023396
Intelligence Index 16.7 43.413.313.457.6
Spring Prompt overall – 40––82
Rank among available models, by use case
Product listings – 20th––13th
Decks from an analysis – 18th––7th
User surveys – 7th––4th
Marketing planning – 30th––9th
Agents and tool use 105th of 159 28th111th131st3rd
Professional work – –––3rd
Reasoning and knowledge 110th of 187 25th116th117th1st
Writing 82nd of 169 –112th154th5th
Sticking to the facts 43rd of 139 ––70th3rd
Speed under pressure – 21st–––

Shaded figures are better than Gemma 4 26B A4B's. A dash means no result.

Where it has been measured

3rd cheapest of the 31 models with similar intelligence, and 3rd cheapest of the 21 models writing 50–100 tokens a second. 8th most intelligent of the 41 models priced under $0.25 per million tokens. Among models available today, mid-field on every use case measured so far.

Every result

37 published results from 4 sources, each in the source's own units, with the configuration that produced it.

Arena (formerly LMArena) · reported by Arena (formerly LMArena) · 7 results
MeasureValueRankConfigurationDated
Arena Text factuality: overallArena rating, higher is better 1,445
1,438–1,452
82 of 182 Gemma 4 26B A4B 2 Oct 2026
Arena Text: business, management and financeArena rating, higher is better 1,444
1,427–1,461
90 of 406 Gemma 4 26B A4B 2 Oct 2026
Arena Text: creative writingArena rating, higher is better 1,400
1,381–1,418
111 of 411 Gemma 4 26B A4B 2 Oct 2026
Arena Text: expert promptsArena rating, higher is better 1,475
1,450–1,501
81 of 363 Gemma 4 26B A4B 2 Oct 2026
Arena Text: instruction followingArena rating, higher is better 1,437
1,424–1,451
83 of 413 Gemma 4 26B A4B 2 Oct 2026
Arena Text: overallArena rating, higher is better 1,437
1,430–1,445
100 of 413 Gemma 4 26B A4B 2 Oct 2026
Arena Text: writing, literature and languageArena rating, higher is better 1,406
1,391–1,421
114 of 412 Gemma 4 26B A4B 2 Oct 2026

Not shown: Not an error rate: claims that cannot be checked on the web are skipped, and preference still carries most of the weight.

Artificial Analysis · reported by Artificial Analysis · 22 results
MeasureValueRankConfigurationDated
Artificial Analysis: Artificial Analysis Coding Indexindex score, higher is better 39.3 122 of 187 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 16.7 238 of 413 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 13.1 276 of 413 Gemma 4 26B A4B (no reasoning) 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 79.2% 179 of 360 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 71.4% 241 of 360 Gemma 4 26B A4B (no reasoning) 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 19.3% 207 of 419 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 11.5% 262 of 419 Gemma 4 26B A4B (no reasoning) 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 72.4% 44 of 288 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 45.4% 165 of 288 Gemma 4 26B A4B (no reasoning) 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 65.7% 227 of 407 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 42.3% 310 of 407 Gemma 4 26B A4B (no reasoning) 20 Oct 2026
Artificial Analysis: Output speedtokens per second, higher is better 53.1 112 of 137 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Output speedtokens per second, higher is better 81.8 82 of 137 Gemma 4 26B A4B (no reasoning) 20 Oct 2026
Artificial Analysis: SciCode% of problems, higher is better 40.0% 163 of 205 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Terminal-Bench 2.1% of tasks, higher is better 39.0% 124 of 188 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 13.6% 181 of 282 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 25.0% 128 of 282 Gemma 4 26B A4B (no reasoning) 20 Oct 2026
Artificial Analysis: Time to first answer tokenseconds, lower is better 38.4 104 of 137 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: Time to first answer tokenseconds, lower is better 0.60 11 of 137 Gemma 4 26B A4B (no reasoning) 20 Oct 2026
Artificial Analysis: τ-bench banking% of tasks, higher is better 12.0% 116 of 175 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 43.6% 174 of 286 Gemma 4 26B A4B 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 40.4% 182 of 286 Gemma 4 26B A4B (no reasoning) 20 Oct 2026

Not shown: Not your codebase or tools.

UGI Leaderboard · reported by UGI Leaderboard · 6 results
MeasureValueRankConfigurationDated
UGI: requested-length error% off the requested word count, lower is better 8.0% 63 of 379 Gemma 4 26B A4B 12 Apr 2026
UGI: requested-length error% off the requested word count, lower is better 8.0% 63 of 379 Gemma 4 26B A4B (channel-thought-prefill reasoning) 12 Apr 2026
UGI: style adherencescore from 0 to 1, higher is better 0.34 224 of 379 Gemma 4 26B A4B 12 Apr 2026
UGI: style adherencescore from 0 to 1, higher is better 0.35 175 of 379 Gemma 4 26B A4B (channel-thought-prefill reasoning) 12 Apr 2026
UGI: writing scorescore out of 100, higher is better 41.6 213 of 379 Gemma 4 26B A4B 12 Apr 2026
UGI: writing scorescore out of 100, higher is better 43.8 196 of 379 Gemma 4 26B A4B (channel-thought-prefill reasoning) 12 Apr 2026

Not shown: Not other format limits such as character counts or bullet counts.

Vectara Hallucination Leaderboard · reported by Vectara Hallucination Leaderboard · 2 results
MeasureValueRankConfigurationDated
Vectara: answer rate% of documents, higher is better 99.8% 34 of 108 Gemma 4 26B A4B 22 Sep 2026
Vectara: hallucination rate% of summaries, lower is better 5.2% 13 of 108 Gemma 4 26B A4B 22 Sep 2026

Not shown: Not a quality score: a low rate usually means content filters were triggered, and hallucination rates are measured on answered documents only.

Will Gemma 4 26B A4B work for your product?

Benchmarks aren't your task. Paste the prompt your product uses and run it beside the models you could use instead, with every answer checked and costed.

New here? Your email starts a free trial. No credit card needed.