feat(opencode): add Moonshot AI Kimi K2.5 model#10835
Conversation
|
Thanks for your contribution! This PR doesn't have a linked issue. All PRs must reference an existing issue. Please:
See CONTRIBUTING.md for details. |
|
The following comment was made by an LLM, it may be inaccurate: No duplicate PRs found |
|
Resolves #10811 |
|
thank you @MPavleski for your work |
|
@palmamartin @vvarp @ndarilek @rmoriz please have it merged fast. it is a good model |
| "last_updated": "2026-01-27", | ||
| "modalities": { "input": ["text", "image", "video"], "output": ["text"] }, | ||
| "open_weights": true, | ||
| "cost": { "input": 0.6, "output": 2.5, "cache_read": 0.15 }, |
There was a problem hiding this comment.
output is 3, not 2.5
https://platform.moonshot.ai/docs/pricing/chat#multi-modal-model-kimi-k25
There was a problem hiding this comment.
Good catch, fixed it, Kimi K2.5 confirmed pricing from screencap:
Thinking: The user has provided a screenshot of the pricing for kimi-k2.5. Let me compare it with what's in the JSON file:
From the image:
- Model: kimi-k2.5
- Input Price (Cache Hit): $0.10 per 1M tokens
- Input Price (Cache Miss): $0.60 per 1M tokens
- Output Price: $3.00 per 1M tokens
- Context Window: 262,144 tokens
Current JSON configuration:
- cost.input: 0.6 (matches Cache Miss price)
- cost.output: 3 (matches Output price)
- cost.cache_read: 0.1 (matches Cache Hit price)
- limit.context: 262144 (matches Context Window)
All the prices match perfectly! The configuration is correct.
All prices match perfectly:
| Metric | Screenshot | Current JSON | Status |
|--------|-----------|--------------|--------|
| Input (Cache Miss) | $0.60 | 0.6 | ✓ |
| Input (Cache Hit) | $0.10 | 0.1 (cache_read) | ✓ |
| Output | $3.00 | 3 | ✓ |
| Context Window | 262,144 | 262144 | ✓ |
The kimi-k2.5 configuration is correct and up to date.
|
@ShivamB25 It's rude to at-mention people randomly to make them do your work. I have no affiliation with this project other than contributing a few PRs so you're just spamming. |
|
@ndarilek appolgies was just excited for model. |
|
What's the excitement? This is just a Chinese model |
It beats Claude Sonnet 4.5 for a fraction of the price. |
|
Is this going to be approved and merged? |
|
I'm closing this down, it is already available on the dev branch / npm repository version,. |
7 participants
What does this PR do?
Adds support for the new Moonshot AI Kimi K2.5 model to both the China (
moonshot-ai-china) and international (moonshot-ai) providers.How did you verify your code works?
Changes:
kimi-k2.5model with full capabilities (attachments, reasoning, tool calls, temperature control)Tested by running
bun devand verifying the model appears in the model list.Also this message was written by Kimi K2.5 based on the commit.
Resolves #10811