IBM Granite 4.2 8B reaches OpenRouter with a 131K context window
OpenRouter added a paid route for IBM Granite 4.2 8B on 31 August. The hosted endpoint exposes a 131,072-token context window, reasoning and tool calling, six days after IBM released the open model family.
Synthesized via Fish Audio S2.1 Pro in British English. Natural editorial summary, not verbatim reading.
Key Takeaways
- check_circleIBM released the Granite 4.2 family on 25 August 2026 in 3B, 8B and 30B sizes under the Apache 2.0 licence.
- check_circleAIMI first saw the OpenRouter route at 22:31:57 SAST on 31 August, six days after IBM's model release.
- check_circleOpenRouter lists a 131,072-token context, 117,964 maximum output tokens, reasoning, tools, function calling and structured outputs for the paid route.
- check_circleThe OpenRouter addition is a hosted access event, not a new IBM announcement. The model weights and IBM release materials were already public.
IBM released Granite 4.2 first
IBM published its Granite 4.2 technical article on 25 August 2026. The family includes dense 3B, 8B and 30B reasoning models, all released under Apache 2.0. The 8B model card identifies IBM's Granite Team as the developer and lists Granite 4.1 8B Base as its starting point.
Granite 4.2 adds built-in thinking, non-thinking and low-effort modes. IBM also documents native tool calling and agentic training for the 8B and 30B models. The model card lists a native 128K context and says the family can be extended to longer contexts.
OpenRouter added the managed API route
OpenRouter added `ibm-granite/granite-4.2-8b` as a paid route. AIMI records provider metadata creation at 22:06:20 SAST and first endpoint observation at 22:31:57 SAST on 31 August. Those timestamps describe OpenRouter availability, not the IBM release date.
The route page lists $0.06 per million input tokens and $0.25 per million output tokens. It exposes a 131,072-token context window and a maximum output of 117,964 tokens. AIMI records reasoning, tools, function calling and structured outputs for this route.
Why the distinction matters
This is a provider-route addition rather than a second model launch. IBM's weights, model card and technical explanation were already available from the maker. OpenRouter adds a managed OpenAI-compatible endpoint, which can be useful when a team wants one API surface across several models.
The route appeared on 31 August without another verified release in this batch. There is no primary evidence that the OpenRouter addition was a response to another company or model. Treat the timing as availability information, not a claim about market causation.
What to test before using it
Test thinking-mode behaviour, tool-call formatting, long-context retrieval and output limits on the exact route your application will use. Provider metadata is a useful starting point, but it is not a substitute for a task-level smoke test.
Granite 4.2 is Apache 2.0 licensed at the model level. The licence does not make every hosted route identical, so check OpenRouter's current pricing, terms and data handling before sending sensitive material.
Access through the AZ Labs AI Gateway
The exact OpenRouter identifier is `ibm-granite/granite-4.2-8b`. AZ Labs customers can use the AI Gateway to keep application code separate from the model route, then compare Granite 4.2 with other providers against their own latency, quality and cost requirements.
Frequently Asked Questions
Is this a new IBM model release on 31 August?
No. IBM released Granite 4.2 on 25 August 2026. The 31 August event was OpenRouter adding a hosted route for the 8B model.
What is the exact OpenRouter route?
The exact route is ibm-granite/granite-4.2-8b. OpenRouter lists it as a paid route at $0.06 per million input tokens and $0.25 per million output tokens.
What capabilities does the route advertise?
AIMI records a 131,072-token context, reasoning, tools, function calling and structured outputs. Test those behaviours on your own integration before relying on them in production.
Is Granite 4.2 open source?
IBM publishes the Granite 4.2 model materials and weights under the Apache 2.0 licence. Hosted provider terms and pricing remain separate from the model licence.
Related Frontier Models & Releases
Explore verified specifications, benchmark results, and route pricing across alternative models in this class.
DeepSeek releases V4.1 Flash with native multimodal vision, 552B MoE and lower API rates
DeepSeek has launched DeepSeek-V4.1-Flash, a 552B-parameter mixture-of-experts model featuring a novel Causal Encoder–Decoder architecture with just 8B active input and 16B active output parameters. The release brings native vision, compresses KV cache storage by up to 8x, and slashes off-peak API pricing to $0.15 per million input tokens.
OpenAI Releases GPT-6 Astra: Next-Generation Flagship with 1.05M Context and Deep Multimodal Reasoning
OpenAI has officially launched GPT-6 Astra, its frontier flagship model featuring a 1,050,000-token context window, 128,000 max output tokens, and native tool-use for autonomous agent workflows.
Alibaba Qwen Releases Qwen3.8 Max: 2.4-Trillion Parameter Flagship with 1M Multimodal Context
Alibaba's Qwen team has launched Qwen3.8 Max, a 2.4T parameter Mixture-of-Experts model offering 1M token context, native video perception, and deep agent tool orchestration.