Side by side
| Model | AA index | Tokens a second | Per 1M tokens: input · cache read · output | An hour of agent work |
|---|---|---|---|---|
| MiMo-V2.6-Pro | 46 | 47 | $0.435 · $0.0036 · $0.87 | $0.51 |
| GLM-5.3 | 45 | 67 | $0.079 · $0.078 · $4.80 | $1.38 |
| Kimi K3 | 44 | 43 | $0.304 · $0.21 · $11.588 | $3.57 |
| GLM-5.3 Flash | 42 | 66 | $0.15 · $0.03 · $0.50 | $0.41 |
| DeepSeek V4.1 Flash | 39 | 179 | $0.018 · $0.0054 · $0.684 | $0.16 |
| For scale: Claude Sonnet 5.5 | 56 | Not open | $2 · $0.10 · $10 | $4.27 |
| Uno AI: Fast and Smart | Two of the models above, behind one key, with web search and dictation | Counted in hours, not tokens | ||
Artificial Analysis Intelligence Index
Higher is better. Open models in grey, the strongest closed models for scale. Checked 16 October 2026.
Show as a table
| Option | index | Shape |
|---|---|---|
| Claude Opus 5.5 | 58 | Closed. The top of the index today |
| GPT-6.1 Sol | 52 | Closed |
| MiMo-V2.6-Pro | 46 | Open weights. Today behind Uno's Smart |
| GLM-5.3 | 45 | Open weights |
| Kimi K3 | 44 | Open weights |
| GLM-5.3 Flash | 42 | Open weights. Today behind Uno's Fast |
| DeepSeek V4.1 Flash | 39 | Open weights. The fastest of the five |
What an hour of agent work costs
Token prices alone hide the bill, because an agent sends its whole context on every call. We counted one hour the way we did for Uno AI against OpenRouter: 150 calls, 50,000 tokens of context each, 45,000 of them from the cache, 1,000 written back, plus OpenRouter's 5.5% fee.
An hour of agent work, our estimate
Same work for every model. OpenRouter list prices on 16 October 2026 plus the fee. Lower is cheaper.
Show as a table
| Option | $ an hour | Shape |
|---|---|---|
| DeepSeek V4.1 Flash | 0.16 | Cheap input and cache reads |
| GLM-5.3 Flash | 0.41 | Today behind Uno's Fast |
| MiMo-V2.6-Pro | 0.51 | Today behind Uno's Smart |
| GLM-5.3 | 1.38 | Cache reads cost nearly as much as new input |
| Kimi K3 | 3.57 | Output is the expensive part |
| Claude Sonnet 5.5 | 4.27 | Closed, for scale |
One caveat: in our own test in September a fast model made more calls a minute, and its real hour cost about twice this estimate. A model that answers faster makes more calls in the same hour, so the cheapest model per token is not always the cheapest per task.
Pick by task
A good habit: plan with the strong model, then hand the steps to a fast one. All five have a context of about 1 million tokens for all five, so a large project fits.
Open weights: can you run them yourself?
Artificial Analysis marks MiMo-V2.6-Pro, GLM-5.3, GLM-5.3 Flash, Kimi K3 and DeepSeek V4.1 Flash as open weights: open weights means you can download and run the model yourself; running a model of this size needs data-center GPUs. For one person or a small team, renting the model through a provider costs far less than the GPUs. That is what OpenRouter does per token, and what Uno AI does per hour.
Where open models and Uno are worse
- The strongest closed models score higher: Claude Opus 5.5 has 58, the best open model 46.
- Claude Code and Codex are tuned for their own vendors' models. With an open model, their tool calls can fail more often.
- Uno AI gives you two of these models, not all five. Others are billed apart: Claude, GPT, Gemini per token from premium credit ($15 Pro, $25 Max) or a top-up.
- On Small+AI and Plus+AI the hours end and the work stops. Only Max+AI is unlimited.
- Scores, speeds and prices change every few weeks, and the models behind Fast and Smart can change too. The names above are today's.
Two open models behind one key, paid in hours
Uno AI writes code, builds sites and runs tasks right on your computer. Only the minutes it works count; ten chats in the same minute count as one.
Today: GLM-5.3 Flash (Fast) and MiMo v2.6 Pro (Smart). Dictation into text at no extra cost, within a daily limit.
- 40 hSmall + AIa month
- 100 hPlus + AIa month
- 45 h + FastPro + AIfull speed, then Fast never stops
- UnlimitedMax + AI100 h at full speed, then standard
Pick a computer, then add Uno AI
Every plan is a real Linux computer with root. Add Uno AI for a fast and a smart model, or sign in with your own Claude or ChatGPT.
Small
A Telegram bot, a VPN, a small backend
$5/ month
- 1 vCPU · 1 GB
- 15 GB disk
- 25 GB cloud storage
- 5 boosts a month
Plus
Agents that work with your laptop closed
$20/ month
- 4 vCPU · 4 GB
- 40 GB disk
- 100 GB cloud storage
- Uno Work in the cloud
Pro
A product in production, a second MacBook
$70/ month
- 16 vCPU · 16 GB
- 120 GB disk
- 300 GB cloud storage
- Speed of a MacBook Air M4
Max
A team and many agents at once
$200/ month
- 32 vCPU · 48 GB
- 300 GB disk
- 1,024 GB cloud storage
- 5 public IPv4
Add Uno AI to any plan: 40 h of AI work a month on Small, 100 h on Plus, unlimited on Max. Today: GLM-5.3 Flash and MiMo v2.6 Pro. Or sign in with your own Claude or ChatGPT.