ibm-granite/granite-4.2-8b

模型发布 OpenRouter 模型 原文语言:中文

OpenRouter 于 2026-09-01 收录该模型(收录时间不等于模型正式发布日期);上下文:131,072 tokens;输入价:$0.0600/百万 tokens;输出价:$0.1500/百万 tokens;定位(来源描述):Granite 4.2 8B is a dense reasoning model from IBM.

全文

IBM: Granite 4.2 8B

ibm-granite/granite-4.2-8b

Model weights

Compare

Granite 4.2 8B is a dense reasoning model from IBM. It is suited for mathematics, code generation, multilingual dialogue, and agentic workflows that need multi-step reasoning. It supports full, low-effort, and non-thinking modes.

The model supports 12 languages: English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, and Chinese.

Modalities

In / out price

$0.06 / $0.15per 1M

Context

131K

Released

Aug 31, 2026

Compare

Providers

Different companies host the same model. OpenRouter routes your request to one of them based on the routing mode you pick: Balanced (price + speed), Nitro (fastest), or Floor (cheapest).

Latency / throughput

$0.06$0.15$0.0150.24s111 tok/s

96.99%

$0.06$0.25$0.0150.30s138 tok/s

100.00%

ProviderInput / Output /M

Input $0.06 / Output $0.15

111 tok/s96.99% uptime

Input $0.06 / Output $0.25

138 tok/s100.00% uptime

Pricing

The average price customers actually pay for this model, next to the prices providers post. Caching and discounts mean the price actually paid is often well below the listed one.

Weighted average

Input

$0.02238/M tokens

Output

$0.1802/M tokens

Granite 4.2 8B — Price history

Price history

No effective price data for this selection

$0.0221$0.14883.8%78.4%

$0.0234$0.24980.8%21.6%

ProviderInput / Output /M

Token share

Effective in $0.0221 / Effective out $0.148

Token share

78.4%

Effective in $0.0234 / Effective out $0.249

Token share

21.6%

Performance

Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).

Throughput

138tok/s

P50, best across providers

Latency

0.24s

P50, best provider

Reasoning effortLatency / throughput

Uptime

Uptime is the percentage of the past 3 days that at least one provider was responding to requests. Availability is the percentage of time that inference was successfully served. OpenRouter continuously monitors and uses the next-best provider when one returns an error.

Uptime (3d)The model was reachable. Request routed to a provider.

100.00%

Availability (3d)The model returned inference from any provider. Errors and empty responses count against it.

98.08%

Availability over the last 3 days

Last 72 hours

Availability 98.08%

3 Days Ago2 Days AgoYesterdayNow

Availability over the last 24 hours

OpenRouter Availability

96.52%

Without Routing

96.49%

When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it. You can access per-provider uptime data programmatically through the Endpoints API. Learn more about our load balancing and provider routing options.

Benchmarks

Scores on standardized evaluations and this model’s own outputs on shared benchmark prompts. Higher percentages are better — and each rank shows where this model lands among the models evaluated on that source.

Benchmark score summary for IBM: Granite 4.2 8B (Artificial Analysis)

SourceBenchmarkScore

Artificial AnalysisGranite 4.2 8B Intelligence Index5.1

Artificial AnalysisGranite 4.2 8B Coding Index22.4

Artificial AnalysisGranite 4.2 8B Agentic Index1.3

Artificial AnalysisGranite 4.2 8B GPQA Diamond63.1%

Artificial AnalysisGranite 4.2 8B HLE9.7%

Artificial AnalysisGranite 4.2 8B AA-LCR45.0%

Artificial AnalysisGranite 4.2 8B τ-Bench Banking7.6%

Artificial AnalysisGranite 4.2 8B GDPval-AA0.0%

Artificial AnalysisGranite 4.2 8B CritPt0.3%

Artificial AnalysisGranite 4.2 8B SciCode31.5%

Artificial AnalysisGranite 4.2 8B Terminal-Bench 4.00.0%

Artificial AnalysisGranite 4.2 8B Terminal-Bench 2.118.4%

Artificial AnalysisGranite 4.2 8B AA-Omniscience Accuracy11.8%

Artificial AnalysisGranite 4.2 8B AA-Omniscience Non-Hallucination Rate67.8%

Apps

Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.

1.

ESCRIBA-01-BibliotecaSV

new

2.45Btokens

2.

Tolokaforge Evaluation

new

1.78Btokens

3.

Kilo Code

Kilo Code is an open-source AI coding agent that works across VS Code, JetBrains, and CLI to help developers ship code faster with agentic workflows.

709Mtokens

4.

GPE Model Gateway

new

277Mtokens

5.

JaimeClaude

new

137Mtokens

Rest of OpenRouter

147M

Top 5 Apps

200

Activity

Token volume and request traffic to this model over time.

API

Drop-in code to call this model. OpenRouter's API is OpenAI-compatible — most SDKs work by just swapping the base URL. The only thing that changes between models is the model slug below.

Not currently available through EU in-region routing. In-region routing docs

Frequently asked questions

Granite 4.2 8B is a dense reasoning model from IBM. It is suited for mathematics, code generation, multilingual dialogue, and agentic workflows that need multi-step reasoning. It supports full, low-effort, and non-thinking modes. The model supports 12 languages: English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, and Chinese.

Granite 4.2 8B costs $0.06/M input tokens and $0.15/M output tokens, with separate rates for Cache Read at $0.015/M tokens.

Granite 4.2 8B has a 131,072 token context window. It supports up to 131,072 completion tokens.

Yes. Granite 4.2 8B accepts tools for function calling. It also supports structured outputs via a JSON schema in response_format.

Granite 4.2 8B is served by 2 providers on OpenRouter: CoreWeave and DeepInfra. Requests are routed to the best available provider, with automatic failover to the others, and you can pin or exclude providers with provider routing.

Granite 4.2 8B was released on August 31, 2026.

正文以来源语言显示。

参考来源

← 返回每日简报