Gemma 4 26B A4B vs Gemma 4 31B API pricing
Updated
Gemma 4 26B A4B: USD 0.07 input, USD 0.34 output per 1M tokens (DeepInfra and 1 other). Gemma 4 31B: USD 0.10 input, USD 0.34 output per 1M tokens (CoreWeave Forge). At 3 input tokens per output token, Gemma 4 31B costs 16% more than Gemma 4 26B A4B (USD 0.16 vs USD 0.1375 per 1M).
How do Gemma 4 26B A4B and Gemma 4 31B differ on price?
- Cheapest input price: Gemma 4 26B A4B USD 0.07 against USD 0.10 per 1M tokens, so Gemma 4 31B costs 1.43× as much (43% more).
- Cheapest output price: Gemma 4 26B A4B USD 0.30 against USD 0.34 per 1M tokens, so Gemma 4 31B costs 1.13× as much (13% more).
- Among the 201 indexed models, Gemma 4 26B A4B ranks 33rd cheapest and Gemma 4 31B 49th on the 3:1 blend.
- No input:output mix changes the answer: Gemma 4 26B A4B is cheaper at every mix before cache discounts.
- Both list cached-input discounts: Gemma 4 26B A4B at 6 standard prices of 11, from USD 0.015 (Google Vertex AI, 90% below that provider's input price); Gemma 4 31B at 6 standard prices of 14, from USD 0.06 (Parasail, 60% below that provider's input price).
- Both state a context window of up to 262,144 tokens.
- Gemma 4 31B has 14 providers with a standard price, 3 more than Gemma 4 26B A4B; 9 providers sell both (CoreWeave Forge, Featherless, Venice, DeepInfra and 5 more).
- On one account, CoreWeave Forge charges USD 0.15 for Gemma 4 26B A4B (1.09× its cheapest price) and USD 0.16 for Gemma 4 31B (its cheapest price) at 3:1, the lowest combined price among the providers that sell both.
- Shopping around matters more for Gemma 4 31B: its dearest standard price is 5.1× its cheapest, against 1.91× for Gemma 4 26B A4B.
- Gemma 4 26B A4B has no batch price on file; Gemma 4 31B has a batch price at 1 provider, from USD 0.09 at 3:1 (Doubleword).
- For the one-developer coding agent preset, the gap is USD 610.20 a year in Gemma 4 26B A4B's favour (Gemma 4 26B A4B USD 117.45 a month, Gemma 4 31B USD 168.30).
- Per character of input, Gemma 4 26B A4B costs 30% less than Gemma 4 31B in every measured language, because both use the Gemma 4 VL 26B tokenizer.
Gemma 4 26B A4B and Gemma 4 31B prices side by side
USD per 1M tokens, standard tier. "Cheapest" is the lowest price at 3 input tokens to 1 output token; each line can come from a different provider.
| Price | Gemma 4 26B A4B | Gemma 4 31B |
|---|---|---|
| Model page | Gemma 4 26B A4B Google · 11 providers | Gemma 4 31B Google · 14 providers |
| Cheapest standard price, blended 3:1 | USD 0.1375 lower USD 0.07 in / USD 0.34 out at DeepInfra and 1 other DeepInfra page checked 2026-11-14 | USD 0.16 USD 0.10 in / USD 0.34 out at CoreWeave Forge CoreWeave Forge page checked 2026-11-15 |
| Cheapest input | USD 0.07 lower DeepInfra source | USD 0.10 CoreWeave Forge source |
| Cheapest cached input | USD 0.015 lower Google Vertex AI source | USD 0.06 Parasail source |
| Cheapest output | USD 0.30 lower Cloudflare Workers AI source | USD 0.34 CoreWeave Forge source |
| Maker's own API | Google publishes no paid standard price on its own API that we found. | Google publishes no paid standard price on its own API that we found. |
| Median standard price | USD 0.13 / USD 0.40 input / output, 11 prices | USD 0.14 / USD 0.40 input / output, 14 prices |
| Context window | 262,144 tokens, largest stated; from 32,768 at some providers | 262,144 tokens, largest stated; from 32,768 at some providers |
| Providers with a standard price | 11 | 14 |
- Lower marks the lower price on a line; tie, equal prices. Other tiers and free rows are on the model pages.
Report a wrong price or number (email to info@broadnet.ai, with this page's link filled in).
What do Gemma 4 26B A4B and Gemma 4 31B cost for a real workload?
30 days at each model's cheapest standard price for the workload (presets of the cost calculator; cached input at the cached price where listed).
| Workload | Gemma 4 26B A4B | Gemma 4 31B | Difference |
|---|---|---|---|
| Support chatbot 2,000 requests a day, 1,500 input tokens (60% cached), 300 output | USD 11.34 Featherless | USD 15.12 CoreWeave Forge | USD 3.78 Gemma 4 26B A4B lower |
| RAG over documents 1,000 requests a day, 8,000 input tokens (30% cached), 500 output | USD 20.46 Featherless | USD 29.10 CoreWeave Forge | USD 8.64 Gemma 4 26B A4B lower |
| Coding agent (one developer) 1,500 requests a day, 40,000 input tokens (85% cached), 2,000 output incl. reasoning | USD 117.45 Google Vertex AI | USD 168.30 Parasail | USD 50.85 Gemma 4 26B A4B lower |
Gemma 4 26B A4B is cheaper on all three workloads. The widest gap is on Coding agent (one developer): the dearer model costs 43% more than the cheaper one.
Does the cheaper model change with your input:output ratio?
Gemma 4 26B A4B is cheaper at every mix of input and output tokens, before cache discounts: its cheapest input price and its cheapest output price are both at or below Gemma 4 31B's, so the input:output ratio never changes the answer.
| Input : output tokens | Gemma 4 26B A4B | Gemma 4 31B |
|---|---|---|
| 1 : 3 (long answers) | USD 0.25 lower Cloudflare Workers AI | USD 0.28 CoreWeave Forge |
| 1 : 1 | USD 0.20 lower Cloudflare Workers AI | USD 0.22 CoreWeave Forge |
| 3 : 1 (the site’s blend) | USD 0.1375 lower DeepInfra | USD 0.16 CoreWeave Forge |
| 10 : 1 (retrieval, long prompts) | USD 0.09455 lower DeepInfra | USD 0.1218 CoreWeave Forge |
| 30 : 1 (agents, big contexts) | USD 0.07871 lower DeepInfra | USD 0.1077 CoreWeave Forge |
USD per 1M tokens at the cheapest standard provider for each mix, before cache discounts. Thinking tokens bill as output.
What do Gemma 4 26B A4B and Gemma 4 31B cost per character in other languages?
Cost to send 1M characters at each model's cheapest non-promotional standard input price, with the Gemma 4 VL 26B tokenizer for Gemma 4 26B A4B (the same one for both, so the gap equals the price gap):
| Content | Gemma 4 26B A4B | Gemma 4 31B |
|---|---|---|
| English | USD 0.015 lower DeepInfra | USD 0.022 CoreWeave Forge |
| Arabic MSA | USD 0.024 lower DeepInfra | USD 0.034 CoreWeave Forge |
| Egyptian Arabic | USD 0.024 lower DeepInfra | USD 0.034 CoreWeave Forge |
| Hindi | USD 0.019 lower DeepInfra | USD 0.027 CoreWeave Forge |
| Chinese | USD 0.047 lower DeepInfra | USD 0.067 CoreWeave Forge |
| Spanish | USD 0.016 lower DeepInfra | USD 0.022 CoreWeave Forge |
| Code | USD 0.020 lower DeepInfra | USD 0.029 CoreWeave Forge |
From the real cost by language dataset (measured tokens per character on FLORES-200 and other texts). Your own text will differ.
Published results on the same benchmark
BroadNet did not run these benchmarks. BroadRouter is BroadNet's site. Each row pairs results from one source with the same setting for both models, as published; results from different sources or settings are never put side by side. One benchmark says little about your own workload.
| Benchmark | Gemma 4 26B A4B | Gemma 4 31B | Source and setting |
|---|---|---|---|
| MMLU Pro | 82.6% | 85.2% | Google model card: Gemma 4 26B A4B, Gemma 4 31B, checked 2026-11-16 Instruction-tuned model |
| GPQA Diamond | 82.3% | 84.3% | Google model card: Gemma 4 26B A4B, Gemma 4 31B, checked 2026-11-16 Instruction-tuned model |
| LiveCodeBench v6 | 77.1% | 80.0% | Google model card: Gemma 4 26B A4B, Gemma 4 31B, checked 2026-11-16 Instruction-tuned model |
| AIME 2026 no tools | 88.3% | 89.2% | Google model card: Gemma 4 26B A4B, Gemma 4 31B, checked 2026-11-16 Instruction-tuned model |
| MMMU Pro | 73.8% | 76.9% | Google model card: Gemma 4 26B A4B, Gemma 4 31B, checked 2026-11-16 Instruction-tuned model |
| GPQA Diamond | 0.732 | 0.758 | Epoch AI, checked 2026-11-16 Minimal reasoning setting, mean score on a 0 to 1 scale |
Epoch AI rows: Epoch AI, Capabilities & benchmarking, licensed CC BY 4.0, shown as the mean score on a 0 to 1 scale, rounded to three decimals (hover a value for the exact figure).
All published results for each model, with every setting: Gemma 4 26B A4B · Gemma 4 31B.
Gemma 4 26B A4B on BroadNet Inference
BroadRouter is BroadNet's own site, so BroadNet's price is shown here for reference and is not part of the comparison above. BroadNet Inference serves Gemma 4 26B A4B as broadnet/gemma-4-26b-a4b at USD 0.10 input and USD 0.22 output per 1M tokens (USD 0.13 blended 3:1; the cheapest third-party standard price above is USD 0.1375). On BroadNet it reads up to 32,768 tokens per request and accepts text and image input (BroadNet rate card).
Check it on your own traffic
- Gemma 4 26B A4B API pricing: all 11 providers and other tiers.
- Gemma 4 31B API pricing: all 14 providers and other tiers.
- Cost calculator with Gemma 4 26B A4B and Gemma 4 31B preloaded
- Gemma token counter
- Prompt caching calculator
More comparisons: Gemma 4 26B A4B vs Qwen3.6 35B A3B · Gemma 4 26B A4B vs gpt-oss-20b · Llama 3.3 70B vs Gemma 4 26B A4B · GLM 5.3 Flash vs Gemma 4 26B A4B · DeepSeek V4 Flash 0731 vs Gemma 4 26B A4B · Gemma 4 26B A4B vs DeepSeek V4 Flash · gpt-oss-120b vs Gemma 4 26B A4B · Gemma 4 31B vs Gemma 3 27B · Llama 4 Scout vs Gemma 4 31B · Gemma 4 31B vs Qwen3.8 27B · DeepSeek V4.1 Flash vs Gemma 4 31B · every Gemma 4 26B A4B comparison · every Gemma 4 31B comparison · all 662 price comparisons
Questions
Is Gemma 4 26B A4B cheaper than Gemma 4 31B?
Yes. On a blend of 3 input tokens to 1 output token, Gemma 4 26B A4B costs USD 0.1375 per 1M tokens and Gemma 4 31B USD 0.16, at each model's cheapest standard price. For input alone, Gemma 4 26B A4B is cheaper (USD 0.07 against USD 0.10 per 1M); for output alone, Gemma 4 26B A4B is (USD 0.30 against USD 0.34). This compares price only, not quality.
What do Gemma 4 26B A4B and Gemma 4 31B cost per 1M tokens?
Gemma 4 26B A4B: from USD 0.07 input and USD 0.34 output at DeepInfra and 1 other; median of 11 standard prices USD 0.13 / USD 0.40. Gemma 4 31B: from USD 0.10 input and USD 0.34 output at CoreWeave Forge; median of 14 standard prices USD 0.14 / USD 0.40. Each price is read from the provider's own pricing page; the source is linked in the table.
Which providers sell both Gemma 4 26B A4B and Gemma 4 31B?
9 providers have a standard price for both: CoreWeave Forge (USD 0.15 and USD 0.16 at 3:1), Featherless (USD 0.1375 and USD 0.18 at 3:1), Venice (USD 0.1975 and USD 0.18 at 3:1), DeepInfra (USD 0.1375 and USD 0.25 at 3:1), Amazon Bedrock (USD 0.1975 and USD 0.205 at 3:1), GMI Cloud (USD 0.1975 and USD 0.205 at 3:1), Novita AI (USD 0.1975 and USD 0.205 at 3:1), Parasail (USD 0.1975 and USD 0.2125 at 3:1), SiliconFlow (USD 0.205 and USD 0.8125 at 3:1). The first is the lowest combined price. Buying each model from its own cheapest provider costs USD 0.1375 and USD 0.16.
Which is cheaper for a coding agent or a chatbot, Gemma 4 26B A4B or Gemma 4 31B?
For a coding agent (1,500 requests a day, 40,000 input tokens (85% cached), 2,000 output incl. reasoning): USD 117.45 a month on Gemma 4 26B A4B (Google Vertex AI) against USD 168.30 on Gemma 4 31B (Parasail), a difference of USD 50.85. For a support chatbot (2,000 requests a day, 1,500 input tokens (60% cached), 300 output): USD 11.34 a month on Gemma 4 26B A4B (Featherless) against USD 15.12 on Gemma 4 31B (CoreWeave Forge), a difference of USD 3.78. Each uses the cheapest standard price for that workload, cached input included. Try your own numbers.
At what input:output ratio does the cheaper of Gemma 4 26B A4B and Gemma 4 31B change?
Gemma 4 26B A4B is cheaper at every mix of input and output tokens, before cache discounts: its cheapest input price and its cheapest output price are both at or below Gemma 4 31B's, so the input:output ratio never changes the answer.
Do Gemma 4 26B A4B and Gemma 4 31B have cached input pricing?
Gemma 4 26B A4B from USD 0.015 per 1M cached input tokens (Google Vertex AI; 6 of 11 standard prices list a discount). Gemma 4 31B from USD 0.06 per 1M cached input tokens (Parasail; 6 of 14 standard prices list a discount). Cached tokens are billed at the input price where no cache price is published.
Which has the larger context window, Gemma 4 26B A4B or Gemma 4 31B?
Both go up to 262,144 tokens on the providers that state it.
Can I use Gemma 4 26B A4B on BroadNet Inference?
BroadNet Inference serves Gemma 4 26B A4B as broadnet/gemma-4-26b-a4b at USD 0.10 input and USD 0.22 output per 1M tokens (rate card synced 2026-11-14). BroadRouter is BroadNet's own site, so BroadNet's price is shown for reference and not ranked against the providers above.
Which scores higher on MMLU Pro, Gemma 4 26B A4B or Gemma 4 31B?
Per the Google model cards (Gemma 4 26B A4B, Gemma 4 31B) (checked 2026-11-16), Gemma 4 26B A4B scores 82.6% and Gemma 4 31B scores 85.2% on MMLU Pro (Instruction-tuned model). BroadNet did not run these benchmarks, and one benchmark says little about your own workload. Every result both models share, with sources.
Read as MarkdownAsk ChatGPTAsk ClaudeAsk Perplexity
Sources
- DeepInfra: pricing (checked 2026-11-14)
- Featherless: pricing (checked 2026-11-15)
- Cloudflare Workers AI: pricing (checked 2026-11-14)
- Google Vertex AI: pricing (checked 2026-11-14)
- CoreWeave Forge: pricing (checked 2026-11-15)
- Parasail: pricing (checked 2026-11-14)
- Featherless: pricing (checked 2026-11-15)
- Venice: pricing (checked 2026-11-15)
- DeepInfra: pricing (checked 2026-11-14)
- Amazon Bedrock: pricing (checked 2026-11-14)
- GMI Cloud: pricing (checked 2026-11-15)
- GMI Cloud: pricing (checked 2026-11-15)
- Novita AI: pricing (checked 2026-11-14)
- SiliconFlow: pricing (checked 2026-11-14)
- Doubleword: pricing (checked 2026-11-16)
- Google model card, gemma-4-26b-a4b-it (checked 2026-11-16)
- Google model card, gemma-4-31b-it (checked 2026-11-16)
- Epoch AI benchmark data, CC BY 4.0 (checked 2026-11-16)
- BroadNet Inference rate card, pricing.json (synced 2026-11-14)