VerdictPal
Join

Models / Google

Gemma 4 12B (Non-reasoning)

Gemma · Released 2026-06-10

Google's compact Gemma 4 non-reasoning variant, evaluated by Artificial Analysis on 10 Jun. The card is intentionally light until benchmark rows are imported into the local ledger.

Cheap local baseline for tagging, extraction, and latency-sensitive prompts where explicit reasoning traces are unnecessary.

#44 of 51 on AA Index · current snapshot
13 AA Index · 10 benchmark rows · 1 sourcesPosition is benchmark-specific — not a cross-family or cross-source ranking.
Composite30
Context128K
Input / 1M—
Output / 1M—
Knowledge cutoffopen
Output speed119 t/s
TTFT (default API)1.49s
Statuslightweight
ReasoningCodingAgenticLong context
Reasoning
66
Coding
30
Agentic
32
Long context
31

40average across 4 tested families

Benchmark placements

Where Gemma 4 12B (Non-reasoning) places on each public benchmark source that publishes a row. Rank counts every model with a latest row in the same test — not a universal quality score.

BenchmarkFamilyRankScoreSourceDate
AA output speedPerformance#23/ 69Artificial Analysis2026-07-21
τ³-Bench BankingAgentic#23/ 6432Artificial Analysis2026-07-21
AA time to first tokenPerformance#35/ 69Artificial Analysis2026-07-21
IFBenchReasoning#49/ 5745Artificial Analysis2026-07-21
AA-LCRLong context#60/ 6431Artificial Analysis2026-07-21
Artificial Analysis Intelligence IndexReasoning#60/ 6713Artificial Analysis2026-07-21
Humanity's Last ExamReasoning#60/ 656Artificial Analysis2026-07-21
SciCodeCoding#60/ 6530Artificial Analysis2026-07-21
GPQA DiamondReasoning#64/ 6966Artificial Analysis2026-07-21
Terminal-BenchAgentic#65/ 6911Artificial Analysis2026-07-21
Family context

The three highest-scoring carded models in each capability family. Where Gemma 4 12B (Non-reasoning) shows up, it's highlighted.

ReasoningGemma 4 12B (Non-reasoning) · 66
  1. 1
    Claude Mythos Preview
    95
  2. 2
    GPT-5.6 Sol
    94
  3. 3
    Kimi K3
    94
CodingGemma 4 12B (Non-reasoning) · 30
  1. 1
    GPT 5.5
    96
  2. 2
    Claude Opus 4.8
    96
  3. 3
    Claude Opus 4.6
    96
AgenticGemma 4 12B (Non-reasoning) · 32
  1. 1
    GPT 5.5
    100
  2. 2
    Claude Opus 4.8
    100
  3. 3
    Claude Fable 5
    100
Long contextGemma 4 12B (Non-reasoning) · 31
  1. 1
    Kimi K3
    75
  2. 2
    GPT-5.1
    75
  3. 3
    GPT-5.6 Sol
    74
Newest receipts

Benchmark rows added to the public ledger for Gemma 4 12B (Non-reasoning) in the last 120 days. Older rows live in the full table below.

10 recent
BenchmarkFamilyScoreSourceDays ago
AA-LCRLong context30.67%Artificial Analysis2
AA output speedPerformance119 t/sArtificial Analysis2
AA time to first tokenPerformance1.49sArtificial Analysis2
Artificial Analysis Intelligence IndexReasoning13.2%Artificial Analysis2
GPQA DiamondReasoning66.1%Artificial Analysis2
Humanity's Last ExamReasoning6.2%Artificial Analysis2
IFBenchReasoning45.17%Artificial Analysis2
SciCodeCoding29.7%Artificial Analysis2
τ³-Bench BankingAgentic31.87%Artificial Analysis2
Terminal-BenchAgentic11.36%Artificial Analysis2
Family coverage

A high composite that hides a weak family is a trap. These bars surface the families where this model hasn't been publicly tested, and where it leads.

Reasoning66
Coding30
Agentic32
Long context31
Price vs. performance
0255075100$0.10$0.30$1$3$10$30input USD / 1M tokens · log scale →← intelligence

Hover or tab any dot for name, score, and input price.

AnthropicOpenAIMistralAlibabaDeepSeekGoogleMoonshotMiniMaxxAIMetaZhipuXiaomiMeituanCohere

Full benchmark ledger

Every catalog benchmark for Gemma 4 12B (Non-reasoning). Scores link to the original source; gaps mean no public row exists yet.

10 sourced rows

10 of 32 catalog benchmarks have a sourced row for Gemma 4 12B (Non-reasoning).

Reasoning

Coding

Agentic

Long context

Performance

Sources