impulse

Mellum2 12B A2 5B Instruct

API id Mellum2-12B-A2.5B-Instruct
Input / 1M
$0.05
at W&B Inference
Output / 1M
$0.10
at W&B Inference
Context window
131K
tokens
Max output
131K
tokens per reply
Hosts
1
1 listings
Released
-
date not published

Pricing

List prices at W&B Inference, in US dollars per million tokens.

TierInput / 1MOutput / 1M
Standard$0.05$0.10

What it costs a month

At W&B Inference list prices, 30 days.

  • Chat assistant5,000 requests a day, 2K in / 0.5K out$22.50
  • RAG over documents2,000 requests a day, 8K in / 0.6K out$27.60
  • Coding agent1,000 requests a day, 40K in / 2K out$66.00
  • Overnight batch20,000 requests a day, 4K in / 0.4K out$144
Use your own numbers

Where to run it

1 host serve Mellum2 12B A2 5B Instruct. The comparison is on blended price, 3 parts input to 1 part output.

HostIn / out per 1M
W&B InferenceInference
Mellum2-12B-A2.5B-Instruct
$0.05 / $0.10

Capabilities

Core

  • Streaming
  • Function calling
  • JSON mode
  • Structured outputs

Multimodal

  • Vision
  • Image input
  • Audio
  • Image generation
  • Video

Advanced

  • Reasoning
  • Tool calling
  • MCP
  • Prompt caching
  • Batch API
  • Embeddings
Takes in
Text
Produces
Text

Cheaper alternatives

Current models that cost less and keep most of what this one can do: tools, vision, reasoning, structured output, caching and enough context.

Same model, cheaper host

No host sells it for less than W&B Inference.

Smaller Independent models

No cheaper Independent model keeps its key capabilities.

Other frontier labs

No cheaper match from another frontier lab.

Open-weight picks

  • Qwen3 8B
    Alibaba Qwen · on Novita AI
    $0.035 / $0.138
    Has every key capability
    3% less
  • Gemma 3 4B IT
    Google · on Amazon Bedrock
    $0.04 / $0.08
    Has every key capability
    20% less
  • Llama 3.2 3B
    Meta · on LlamaGate
    $0.04 / $0.08
    Has every key capability
    20% less
  • Granite 4.0 H Micro
    IBM · on Cloudflare Workers AI
    $0.017 / $0.112
    Has every key capability
    35% less

Call it

Using W&B Inference's API.

Python
from litellm import completion

response = completion(
    model="wandb/Mellum2-12B-A2.5B-Instruct",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Quick answers

How much does Mellum2 12B A2 5B Instruct cost?

On W&B Inference, Mellum2 12B A2 5B Instruct costs $0.05 per million input tokens and $0.10 per million output tokens.

Which APIs offer Mellum2 12B A2 5B Instruct?

Mellum2 12B A2 5B Instruct runs on 1 API: W&B Inference.

What is Mellum2 12B A2 5B Instruct's context window?

131K tokens, with up to 131K tokens in one reply.

Is Mellum2 12B A2 5B Instruct being retired?

It has been retired: W&B Inference retired it on 5 Oct 2026.

What are cheaper alternatives to Mellum2 12B A2 5B Instruct?

Models that cost less and keep most of what it can do: Qwen3 8B on Novita AI (3% less). Savings are on a mix of 3 input tokens to 1 output token.

What is Mellum2 12B A2 5B Instruct's API model id?

Use Mellum2-12B-A2.5B-Instruct with W&B Inference.

Sources

Raw catalog entry for Mellum2-12B-A2.5B-Instruct on W&B Inference
{
  "deprecation_date": "2026-10-05",
  "max_output_tokens": 131072,
  "max_input_tokens": 131000,
  "input_cost_per_token": 5e-8,
  "output_cost_per_token": 1e-7,
  "litellm_provider": "wandb",
  "mode": "chat",
  "supports_vision": false,
  "source": "https://wandb.ai/site/pricing/tokens/"
}