Verified 2026-11-18 · rates follow the live catalog
Avian is a prepaid, per-token inference API for ten open-weight models — DeepSeek, MiniMax, GLM, Kimi and MiMo — behind one OpenAI-compatible endpoint. It sells no subscription, credits do not expire, and it states there are no rate limits; it also rents dedicated 8-GPU B200 and B300 nodes by the term. Openference adds plans with bundled usage, team seats and frontier models alongside the open ones; the table below compares per-token rates on the models both list.
Hosted gateway · Pay-as-you-go (prepaid credits) · Checked 19 November 2026 against Avian Pricing.
Pick Avian for prepaid pay-per-token open models with no subscription. Pick Openference for bundled-usage plans, team seats and frontier models next to open ones.
| Plan | Price | What it includes |
|---|---|---|
| Pay-as-you-go | Per token | Prepaid credits, deducted per request. Ten models including DeepSeek V4.1 Flash, MiniMax M3, GLM-5.3 and Kimi K3. OpenAI-compatible endpoint. |
| Dedicated nodes | By the term | Whole 8-GPU B200 or B300 HGX nodes, single-tenant, prepaid for 12 months to 3 years. US only. |
| Avian | Openference | |
|---|---|---|
| Pricing model | Prepaid credits, billed per token | Plan-based (included usage) + on-demand credits |
| Entry point | Prepaid top-ups ($100, $150 and $250 listed) | From $20/mo (Minion) |
| Models included | 10 open-weight models | Open and frontier models |
| Proprietary frontier models | No | Yes |
| Credit expiry | Credits never expire | Plan allowance renews each period |
On the 8 models both list, Avian has the lower price for a typical request on 6 and Openference on 1; 1 model cost about the same. The widest gap in Openference's favour is MiniMax M3 (88% lower); the widest in Avian's favour is DeepSeek V4 Pro (58% lower).
| Model | Avian per 1M (in / out) | Openference per 1M (in / out) | Avian, one request | Openference, one request | Lower price |
|---|---|---|---|---|---|
| GLM-5.3 | $1.19 / $3.74 | $1.40 / $4.40 | $0.028 | $0.032 | Avian, 15% lower |
| GLM-5.3-Flash | $0.1275 / $0.425 | $0.15 / $0.50 | $0.003 | $0.0035 | Avian, 15% lower |
| DeepSeek V4.1 Flash | $0.1275 / $0.51 | $0.30 / $1.20 | $0.0031 | $0.0072 | Avian, 57% lower |
| DeepSeek V4 Pro | $0.561 / $1.683 | $1.32 / $3.96 | $0.013 | $0.03 | Avian, 58% lower |
| MiniMax M3 | $0.255 / $1.02 | $0.03 / $0.12 | $0.0061 | $0.00072 | Openference, 88% lower |
| Kimi K3 | $2.55 / $12.75 | $3 / $15 | $0.064 | $0.075 | Avian, 15% lower |
| MiMo-V2.6-Flash | $0.14 / $0.28 | $0.14 / $0.28 | $0.0031 | $0.0031 | About the same |
| MiMo-V2.6-Pro | $0.36975 / $0.7395 | $0.435 / $0.87 | $0.0081 | $0.0096 | Avian, 15% lower |
A request here is a 20,000-token prompt answered with 1,000 tokens, with no prompt caching. Avian's rates are as published on Avian Pricing, read 19 November 2026. Openference's are read from the live catalog at the highest rate of the day, so a model with a scheduled off-peak discount costs less at other hours than shown. Prices on both sides change; check the sources before you commit.
Pay-as-you-go (prepaid credits). The plan table above lists each tier as Avian publishes it, read from Avian Pricing on 19 November 2026.
On the 8 models both list, Avian has the lower price for a typical request on 6 and Openference on 1; 1 model cost about the same. The table above shows each model; which side is cheaper overall depends on which models you actually run.
Avian's from its own pages on 19 November 2026; Openference's from its live catalog and public plans. Both change, so confirm on the linked page before you commit.
More on the comparison index, the current plans and the model catalog.
Model catalog · Pricing · Comparisons · Cost guides · Integration guides · FAQ
Guides: GLM coding plan alternative · Free AI API · LLM gateway · LLM proxy · Cheap AI API · OpenAI-compatible API
Model APIs: GLM API · DeepSeek API · Kimi API · Qwen API · MiniMax API