[Chatbot Arena] Add GLM-4 variants: AirX, Air and Flash
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 39.5k
- Forks
- 4.8k
- PR merge metrics
- No merged PRs in 30d
Description
Currently GLM-4-0520 is available on the leaderboard and performs really well. However, Zhipu AI also has other variants available, which are 10x, 100x and 1000x as cheap. It would be very interesting to see how they perform.
They are also the only LLM provider that cover a 1000x price range. It could be one of the most interesting data points to see how LLM performance scaled on an (assumed) similar platform.
See https://open.bigmodel.cn/pricing
| Model | Overview | Price(1K tokens) | Est. price 1M tokens in USD |
|---|---|---|---|
| GLM-4-0520 | Our most advanced and intelligent model to date, with an 18.6% improvement in instruction compliance, 128k context, released on 2024-06-05. | ￥0.1 | $ 0.63 |
| GLM-4V | Supports visual QA, image captioning, visual positioning, and complex object detection among other image understanding tasks, with 2k context. | ￥0.05 | $ 0.31 |
| GLM-4-AirX | High-performance version of GLM-4-Air, same effectiveness, 2.6 times faster inference speed. | ￥0.01 | $ 0.063 |
| GLM-4-Air | Best cost-performance model, similar overall performance to GLM-4, with 128k context, fast and affordable. | ￥0.001 | $ 0.0063 |
| GLM-4-Flash | Suitable for simple tasks, fastest speed, most affordable version,with 128k context. | ￥0.0001 | $ 0.00063 |
On Chatbot Arena leaderboard:
- GLM-4-0520
- GLM-4V
- GLM-4-AirX
- GLM-4-Air
- GLM-4-Flash
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No files or tests are named. Start by tracing how GLM-4-0520 is registered in Chatbot Arena and how leaderboard model entries are defined. Add the requested GLM-4V, AirX, Air, and Flash variants, then verify that each appears and can be evaluated on the leaderboard.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai
- Issue type
- Feature
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 42/100