MoE workstation
Stealth model, operator anonymous. Union Alpha appeared on OpenCode on 2026-09-16 as a free one-week preview (“Union Alpha Free”, model id opencode-go/union-alpha) and the same day on OrcaRouter as a free OpenAI-compatible endpoint. OpenCode’s copy: built for agentic coding, supports images, no training on user data, capacity “5T tokens a day so far, potentially can go up”. Asked directly, the model answers that the maker is anonymous and it is a stealth model. No vendor, checkpoint, parameter count, or context window is published.
The GLM hypothesis: community fingerprinting (run by Pliny the Liberator and others) found 21/21 image-bearing probes matching GLM-5.3 Flash prompt-token counts after a constant -17 offset, andgMASK /</think> markers adding +2 tokens in both Union and the GLM control. That is strong GLM-compatible tokenizer evidence, but the exact checkpoint and operator remain unconfirmed - GLM-5.4 is a plausible hypothesis, not an established identity. This card therefore carries no developer attribution. #
Measured performance: Artificial Analysis pegs it around 53% on Terminal-Bench v4.0 at an anticipated per-task cost near $2.50 - above GLM-5.3 (max) and level with GPT-6 Astra (high) on the same chart. If it holds up under replication, that is frontier-adjacent agentic coding at a flash-tier price. #
Availability: free during the launch window on both OpenCode and OrcaRouter (see the cloud-pricing table); a permanent price has not been announced.
Treat every number here as provisional until the operator identifies itself and third parties replicate the benchmarks.
- proprietary
- Sep 2026
Scores #
Save your hardware and every model page answers the real question: will it run on your machine, and how fast?
Or run it in the cloud #
Live per-provider pricing, throughput and uptime. Click a column to sort.
| Provider | Type | Input $/M | Output $/M | Cache $/M | Tok/s | Latency | Uptime | Value |
|---|---|---|---|---|---|---|---|---|
| | Sub | - | - | - | - | - | - | $10.00/mo Go ($5 first month) |
Default order: throughput among 95%+ uptime providers, then latency; subscriptions last. Sort by any column. Subscription rows show $/mo in the Value column - per-token columns are "-". Affiliate links are marked sponsored / nofollow. Confirm current pricing on the provider's site before committing.
Detailed API pricing page + JSON endpoint →
Inference cost over time #
Data accumulates from the first daily sync - longer ranges populate over time. Prices come from OpenRouter snapshots, not a historical API.