What happened
On October 30, 2026, OpenCode merged pull request #60625, titled “docs(go): raise DeepSeek V4.1 Flash limits to $100/$200/$105.” The change reapplies a prior increase that had been reverted. The PR says the Go documentation and the opencode.ai/go page show monthly DeepSeek V4.1 Flash limits of $100 for Go, $200 for Go Plus, and $105 for Go Teams. It also says request estimates were scaled at the same per-dollar rate as before.
The PR describes the change as an update to the Go plan figures and associated limits, not a change to DeepSeek’s underlying model or its token price. A reviewer comment says the estimate remains about 433 requests per dollar and that the estimates agree between the Go model data and the documentation. The PR references prior internal limit values of 69,330 requests/$160 and 36,837/$85, moving to 86,670/$200 and 45,500/$105 respectively. The accompanying intake record summarizes the earlier Go tier as having moved from $80 to $100, and Go Plus from $160 to $200; it also reports the third tier moving from $85 to $105. These are plan allowances, not a promise that every user will receive a fixed count of requests for every workload.
The merged change was primarily documentation and marketing data in the OpenCode repository. The PR states that the Console chart was reapplied in the separate opencode-console repository and that the enforced limits were set in the inference registry. A reviewer noted that the API itself was unchanged and asked that registry limits and the Console change land before the next promotion from development to production, because an earlier revert had been made to prevent advertised limits from exceeding what the registry and Console enforced. The supplied intake material reports that the change shipped in OpenCode v2.2.4 on the same day. The PR’s merge record is dated October 30, 2026.
For a builder, the practical change is more monthly usage budget on those Go tiers for DeepSeek V4.1 Flash. The source does not describe a new model capability, a different context limit, a new API, or a change in the model’s per-token price. Its request estimate is a planning aid; the same budget can cover different numbers of requests depending on how much each request consumes.
What the archive adds
This is a new fact on a closely tracked story, not a first announcement. OpenCode made the $60 DeepSeek V4.1 Flash quota permanent on October 24; two days later, OpenCode raised the Go allowance from $60 to $80. The latest PR moves the Go and Go Plus figures higher again, and adds the $105 Go Teams amount in the available source material.
Trend signal
No supplied SITE_PATTERNS thesis directly describes repeated increases to a hosted model allowance. This item documents a plan-budget change, but it does not establish a broader shift in agent economics or in how teams measure completed work.
Why now
The PR says the higher figures had previously been applied and then reverted because advertised limits risked exceeding the limits enforced by the inference registry and Console. It reapplies the increase and calls out rollout alignment as the remaining operational check. The supplied intake material says the change shipped with OpenCode v2.2.4 on October 30, 2026.
Connections to other events
A prior OpenCode update made DeepSeek V4.1 Flash free on its platform subject to a 200K context cap, a distinct access path from the Go plan allowance. The current PR concerns paid Go-tier monthly limits; it does not say that the free option or its context cap changed. The earlier free-access note therefore provides context, not evidence that the two offers have been combined.
Why it matters
For teams that route coding or agent workloads through OpenCode Go, a larger monthly allowance can reduce the chance that a useful workflow stops when its included budget is exhausted. The increase is especially relevant to teams that already chose DeepSeek V4.1 Flash on the basis of its cost profile: the allowance rises while the per-token rate is described as unchanged. That makes it a budget-capacity change, not a unit-cost improvement.
The operational lesson is to distinguish the published plan amount from actual workload capacity. The PR’s estimate of about 433 requests per dollar cannot predict a team’s request count without knowing prompt size, output length, and task mix. The source also records a previous mismatch between advertised plan figures and enforced limits, so teams with production dependencies should confirm what their account currently exposes rather than changing forecasts solely from a documentation update. For light experimentation, the higher amount may make it easier to test a workflow without immediately purchasing additional capacity; for production planning, teams still need usage monitoring and a fallback when a quota is reached.
Limits and uncertainty
The source is an OpenCode repository pull request, not an independent audit of account-level billing or quota enforcement. It says the documentation and Go page were updated and that the registry limits and Console chart were reapplied, while a reviewer flags rollout timing as a check. The supplied intake record says the change shipped in v2.2.4, but the PR itself does not establish that every user saw the new allowance at the same moment. The material gives no details on eligibility, regional availability, rollover, throttling, overage charges, or the precise workload behind the request estimates. It also does not establish any change to DeepSeek V4.1 Flash’s model quality, latency, context window, or token pricing.
Reader takeaway
OpenCode’s latest documented increase gives DeepSeek V4.1 Flash users higher Go-tier monthly allowances: $100 on Go, $200 on Go Plus, and $105 on Go Teams. Treat it as more included budget at the existing rate, not cheaper tokens or a fixed request guarantee. If the quota matters to a live workflow, verify the limit currently enforced for your account before revising capacity plans.