Compare / Nova 2 Lite vs Gemma 4 26B A4B

Nova 2 Lite vs Gemma 4 26B A4B

Which is cheaper, which is more capable, and which does better at each kind of work, from 23 results on 4 sources that measured both models.

AmazonNova 2 Lite Released 29 Oct 2025 · $0.30 in · $2.50 out
GoogleGemma 4 26B A4B Released 2 Apr 2026 · $0.090 in · $0.30 out
Nova 2 Lite wins 4Gemma 4 26B A4B wins 19

Of the 23 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.

Results as of 16 November 2026 (catalogue release 2026-11-16-49bdfa045dee). See also Amazon vs Google, every model from both labs.

Nova 2 Lite or Gemma 4 26B A4B? The short answer

Pick Gemma 4 26B A4B: it scores higher (Intelligence Index 16.7 against 13.4) and costs less. Pick Nova 2 Lite only if speed matters more: it writes faster.

  • Gemma 4 26B A4B is cheaper: $0.14 against $0.85 per million tokens, 6.0 times less.
  • Gemma 4 26B A4B scores higher on the Artificial Analysis Intelligence Index: 16.7 against 13.4.
  • Nova 2 Lite writes faster: 233 tokens a second against 82.
  • Gemma 4 26B A4B is newer: released 2 Apr 2026, against 29 Oct 2025 for Nova 2 Lite.
  • On third-party benchmarks, on the results both models have: Gemma 4 26B A4B does better for agents and tool use, reasoning and knowledge and writing.

Head to head

13.4
Intelligence Index
16.7
$0.85
Price per million tokens, blendedlower is better
$0.14
233
Output speed, tokens a second
82
1,000,000
Context window, tokens
262,144

Every fact side by side

Nova 2 LiteGemma 4 26B A4B
Price per million tokens$0.30 in · $2.50 out$0.090 in · $0.30 out
Blended (3 in : 1 out)$0.85$0.14
Cheapest host, blended$0.85 (Amazon Bedrock)$0.068 (Darkbloom)
Intelligence Index13.416.7
Spring Prompt overall––
Output speed, tokens a second23382
Context window1,000,000 tokens262,144 tokens
Released29 Oct 20252 Apr 2026 (newer)
WeightsClosedOpen
Developer based inthe USthe US

Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.

Which is better for what

Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.

Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 131 of 159 105 of 159
Reasoning and knowledgeHard questions across science, maths and general knowledge 117 of 187 110 of 187
WritingWhat people prefer in blind comparisons, and judged writing quality 154 of 169 82 of 169
Sticking to the factsSummarising without inventing things, and factual answers 70 of 139 43 of 139

Every shared result

Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.

MeasureNova 2 LiteGemma 4 26B A4B
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 3 results
Overall · TextArena rating, higher is better 214 · 87 of 387 measured 1,335 1,437
Overall · Text factualityArena rating, higher is better 143 · 71 of 163 measured 1,380 1,445
Business, management and finance · TextArena rating, higher is better 186 · 78 of 380 measured 1,351 1,444
Artificial Analysis reported by Artificial Analysis · 3 results
Artificial Analysis Intelligence Indexindex score, higher is better 143 · 123 of 249 measured 13.4high reasoning 16.7
GPQA Diamond% of questions, higher is better 106 · 113 of 232 measured 81.1%high reasoning 79.2%
Humanity's Last Exam% of questions, higher is better 150 · 119 of 250 measured 11.6%high reasoning 19.3%
UGI Leaderboard reported by UGI Leaderboard · 3 results
Writing score · UGIscore out of 100, higher is better 190 · 92 of 230 measured 23.5no reasoning 43.8channel-thought-prefill reasoning
Requested-length error · UGI% off the requested word count, lower is better 200 · 25 of 230 measured 45.0%thinking reasoning 8.0%
Style adherence · UGIscore from 0 to 1, higher is better 108 · 93 of 230 measured 0.35no reasoning 0.35channel-thought-prefill reasoning
Vectara Hallucination Leaderboard reported by Vectara Hallucination Leaderboard · 2 results
Hallucination rate · Vectara% of summaries, lower is better 10 · 13 of 104 measured 5.1% 5.2%
Answer rate · Vectara% of documents, higher is better 45 · 31 of 104 measured 99.6% 99.8%
Show all 23 results (12 more)
MeasureNova 2 LiteGemma 4 26B A4B
Arena (formerly LMArena) reported by Arena (formerly LMArena) · 4 results
Creative writing · TextArena rating, higher is better 248 · 96 of 385 measured 1,269 1,400
Expert prompts · TextArena rating, higher is better 192 · 71 of 338 measured 1,352 1,475
Instruction following · TextArena rating, higher is better 199 · 71 of 387 measured 1,327 1,437
Writing, literature and language · TextArena rating, higher is better 235 · 97 of 386 measured 1,298 1,406
Artificial Analysis reported by Artificial Analysis · 8 results
Terminal-Bench 2.1% of tasks, higher is better 96 · 75 of 131 measured 16.1%high reasoning 39.0%
Artificial Analysis Coding Indexindex score, higher is better 93 · 74 of 130 measured 23.0high reasoning 39.3
IFBench% of instructions, higher is better 50 · 40 of 195 measured 70.7%high reasoning 72.4%
Long-context reasoning (AA-LCR)% of questions, higher is better 147 · 131 of 241 measured 60.3%high reasoning 65.7%
Output speedtokens per second, higher is better 12 · 52 of 78 measured 233no reasoning 81.8no reasoning
Terminal-Bench Hard% of tasks, higher is better 100 · 78 of 190 measured 17.4%medium reasoning 25.0%no reasoning
Time to first answer tokenseconds, lower is better 19 · 11 of 78 measured 0.73no reasoning 0.60no reasoning
τ²-bench telecom% of tasks, higher is better 76 · 110 of 193 measured 75.7%medium reasoning 43.6%

Quick answers

Is Nova 2 Lite better than Gemma 4 26B A4B?

Pick Gemma 4 26B A4B: it scores higher (Intelligence Index 16.7 against 13.4) and costs less. Pick Nova 2 Lite only if speed matters more: it writes faster.

Which is cheaper, Nova 2 Lite or Gemma 4 26B A4B?

Gemma 4 26B A4B is cheaper: $0.14 against $0.85 per million tokens, 6.0 times less.

Which is faster, Nova 2 Lite or Gemma 4 26B A4B?

Nova 2 Lite writes about 233 tokens a second on its usual API, against 82 for Gemma 4 26B A4B (Artificial Analysis).

Which is better for agents and tool use, Nova 2 Lite or Gemma 4 26B A4B?

Gemma 4 26B A4B ranks 105 of 159 models on sale for multi-step tasks with tools: support desks, coding agents, function calls, against 131 for Nova 2 Lite.

Which is better for reasoning and knowledge, Nova 2 Lite or Gemma 4 26B A4B?

Gemma 4 26B A4B ranks 110 of 187 models on sale for hard questions across science, maths and general knowledge, against 117 for Nova 2 Lite.

Which is better for writing, Nova 2 Lite or Gemma 4 26B A4B?

Gemma 4 26B A4B ranks 82 of 169 models on sale for what people prefer in blind comparisons, and judged writing quality, against 154 for Nova 2 Lite.

Which is better for sticking to the facts, Nova 2 Lite or Gemma 4 26B A4B?

Gemma 4 26B A4B ranks 43 of 139 models on sale for summarising without inventing things, and factual answers, against 70 for Nova 2 Lite.

Which has the bigger context window, Nova 2 Lite or Gemma 4 26B A4B?

Nova 2 Lite: 1,000,000 tokens, against 262,144 for Gemma 4 26B A4B.

Run Nova 2 Lite and Gemma 4 26B A4B on your own prompt.

Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.