Granite 4.2 3B
Granite 4.2 3BAlternatives?
CheaperGPT-6 Lunalow×3.1-25%$0.005Function calling only with none · Chat Completions+3IOpenK2 Horizon 7Bthinking×2.7+900%$0.30 / $1.20Memory (4-bit): 1.8 GB → 4.5 GBCheaperGPT OSS 20BlowOpen-20%-31%$0.018 / $0.09Memory (4-bit): 1.8 GB → 13 GB
CheaperQwen3.5 0.8Bnon-thinkingOpen-23%-62%$0.01 / $0.05CheaperGranite 4.0 Micronon-thinkingOpen-24%-22%$0.017 / $0.11GPT-6 Lunalow×3.1-25%$0.005Function calling only with none · Chat Completions+3GPT OSS 20BlowOpen-20%-31%$0.018 / $0.09Memory (4-bit): 1.8 GB → 13 GBQwen3.5 0.8Bnon-thinkingOpen-23%-62%$0.01 / $0.05Granite 4.0 Micronon-thinkingOpen-24%-22%$0.017 / $0.11Qwen3.5 2Bnon-thinkingOpen-25%-24%$0.02 / $0.10GPT-6 Lunalow×3.1-25%$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow×3+64%$0.01GPT OSS 20BhighOpen+66%+91%$0.011Memory (4-bit): 1.8 GB → 13 GB
Mistral Small 4thinkingOpen+49%+151%$0.015Memory (4-bit): 1.8 GB → 72 GB
Nemotron 3 Nano 30B A3BthinkingOpen+72%+180%$0.017Memory (4-bit): 1.8 GB → 19 GB
Gemma 4 12BthinkingOpen×2.4+186%$0.10 / $0.30Memory (4-bit): 1.8 GB → 6 GBQwen 3.5 9BthinkingOpen×2.3+3,451%$0.21Memory (4-bit): 1.8 GB → 4.8 GBIK2 Horizon 3.7BthinkingOpen×2.1+900%$0.30 / $1.20Memory (4-bit): 1.8 GB → 2.5 GBTFalcon H1R 7BthinkingOpen+66%——Memory (4-bit): 1.8 GB → 3.8 GBGranite 4.2 30BthinkingOpen+69%+1,097%$0.072Memory (4-bit): 1.8 GB → 18 GBGranite 4.2 8BthinkingOpen+46%+294%$0.024Memory (4-bit): 1.8 GB → 4.4 GBGranite 4.0 H 350MOpensame——Context: 128K → 33KGranite 4.0 350Mnon-thinkingOpen-17%——Context: 128K → 33KGranite 4.0 H 1BOpen-24%——Granite 4.0 Micronon-thinkingOpen-24%—$0.017 / $0.11Granite 4.0 1Bnon-thinkingOpen-27%——Granite 4.1 30BOpen-38%—$0.10 / $0.25Memory (4-bit): 1.8 GB → 17 GBAll IBM models →Run it yourself
estimateHistory
2 versionsBenchmarks
in Chat & KnowledgeFAQ
How much does Granite 4.2 3B cost?›
$0.03 per 1M input tokens and $0.12 per 1M output tokens (cheapest host: DeepInfra). A typical benchmark task costs about $0.006 at default effort.
Is Granite 4.2 3B free to use?›
There is no free API; the cheapest paid price is $0.03 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Granite 4.2 3B, and what is it best at?›
#230 of 334 in Chat & Knowledge, #75 of 132 in Coding, #124 of 133 in Agents & Tool Use, #174 of 195 in Writing & Language, #130 of 133 in Vision & Documents.
Is Granite 4.2 3B better than Granite 4.1 3B?›
On Humanity's Last Exam, Granite 4.2 3B scores 6.6 vs 3.4 — higher.
How can I use Granite 4.2 3B?›
Through the API of DeepInfra.
What is the context window of Granite 4.2 3B?›
128K tokens.
When was Granite 4.2 3B released? Is there a newer version?›
Released 2026-08-25. It is the current version in its line.
Is Granite 4.2 3B open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Granite 4.2 3B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Granite 4.2 3B read images?›
No, it accepts text only.
What is Granite 4.2 3B?›
Granite 4.2 3B is an open-weights language model from IBM, released Aug 2026, the successor to Granite 4.1 3B.
What are the best alternatives to Granite 4.2 3B?›
GPT-6 Luna (cheaper), K2 Horizon 7B (open weights).