OpenCode Go

opencode.ai · Coding subscriptions, Inference · 2 records · JSON

OpenCode Go Coding subscriptionscaveatcardcardDrequest-rate
OpenCode Zen free models Inferencefails: identityemailno cardM

OpenCode Go Coding subscriptions

caveatcardcardDrequest-rate

opencode.ai/docs/go · coding subscription · coding-subs/opencode-go · JSON · compare

Free tier	Go is USD 10 a month and Go Plus USD 40; three models are Free with an Unlimited monthly limit 'limited time', and free models stay usable after the paid limit is reached unit mixed; metered on request-count; the paid limit is a monthly dollar amount per model, not a token count
Price	none stated
Per-object cap	not applicable: a subscription meters usage, not objects
Retention	not applicable
Egress	not applicable: the vendor serves the inference
Identity	card billing details and a card in the console; one seat per workspace, and a workspace with a Go subscription cannot invite members
Limits	request-rate per-model monthly dollar ceiling; the 5-hour cap is 20 per cent of it, the weekly cap 50 per cent, the month 100 per cent source request-rate no per-request limit is published. The page publishes an estimated request count per model and names the token shape each estimate assumes, for example GLM-5.3/5.2 at 700 input, 52,000 cached and 150 output tokens per request, under 'The estimates use the following token counts per request; actual usage varies' source
Binding	request-rate
Interface	rest, cli an API key from the OpenCode console; endpoints are published per model family at opencode.ai/inference/go/{openai,anthropic,google}/
Tooling	OpenCode TUI, or any client pointed at the published endpoint
Composes with	the only plan found in this topic that publishes both a per-model dollar ceiling and the API endpoint that spends it
Reliability	no reliability event recorded
Notes	The GLM-5.3 cap of USD 15 a month at USD 1.40 per 1M input is about 10.7M input tokens with no output.
Evidence rung	documented the ceilings, the reset windows and the endpoints are quoted from the vendor page read today. The dollar ceiling has not been converted to tokens by measurement. verified_at is the instant its evidence URLs were fetched on 2026-10-22, read from the fetch log, not typed.
Checked	2026-10-23 (live); stale on 2026-11-22

Evidence (2)

date	source	what it proved
2026-10-22	opencode.ai/docs/go	Go USD 10 a month and Go Plus USD 40; usage limits are monthly dollar amounts per model, with 5-hour, weekly and monthly allowances at 20, 50 and 100 per cent; 16 of the 36 Go ceilings are USD 15, 13 are USD 60, two are USD 30, two are USD 80 and three are Unlimited; the page publishes estimated request counts per model against a named token shape and states no per-request limit; exceeded limits fall back to pay-as-you-go credit if enabled, otherwise requests are blocked and the free models remain usable
2026-10-22	opencode.ai/docs/zen	the pay-as-you-go Console prices 14 model ids at Free for input, output and cached read, all marked Free or limited time

Unknown or withheld values (2)

$.reliability_note none-observed
$.retention.stated not-applicable

OpenCode Zen free models Inference

fails: identityemailno cardM

opencode.ai/zen/v1 · inference endpoint, access-gated · inference/opencode-zen-free · JSON · compare

Free tier	the Zen pricing table lists 14 model ids at Free for input, output and cached read, each marked Free or limited time unit unlimited; metered on request-count; 'Free' in the price column is not 'reachable': the free tier is restricted to the OpenCode client
Price	none stated
Per-object cap	not applicable: an inference endpoint stores nothing for the caller
Retention	not applicable
Egress	the response is the payload; no separate egress term is published
Identity	email an OpenCode account, and the request must originate from the OpenCode client. The 403 body names the restriction: 'OpenCode's free tier can only be used from within OpenCode'
Binding	none
Interface	rest endpoint: https://opencode.ai/zen/v1/chat/completions
Tooling	any OpenAI-compatible client; curl against POST {base}/chat/completions
Composes with	no chain exists for a script: the free tier answers only inside the OpenCode client, so the client is the scarce link and it cannot be substituted
Reliability	the restriction is enforced on the request path, not the account, so no header or key makes an external call succeed
Evidence rung	measured four calls across two sessions with two User-Agents. The 403 is a named policy gate and not an outage, and the body says so.
Checked	2026-10-23 (live); stale on 2026-11-22

Evidence (3)

date	source	what it proved
2026-10-22	probe:https://opencode.ai/zen/v1/chat/completions	longcat-2.5-preview-free returned 403 FreeTierError 'OpenCode's free tier can only be used from within OpenCode' in both runs, and with both a curl and a Python User-Agent
2026-10-22	probe:https://opencode.ai/zen/v1/chat/completions	not published
2026-10-22	opencode.ai/docs/zen	14 model ids priced Free for input, output and cached read, including Big Pickle, GLM-5.3-Flash Free, DeepSeek V4.1 Flash Free and LongCat 2.5 Preview Free

Unknown or withheld values (5)

$.evidence[1].proved withheld
$.notes withheld
$.probe.records[0].tps_basis process
$.probe.records[1].tps_basis process
$.retention.stated not-applicable

Corrections

2026-10-22 The free inference endpoint measured on 2026-10-02 was already dead by 2026-10-22, and four more changed state inside one session