explainx / blog
Cline launches a $9.99/month subscription with 2–5× discounted GLM-5.2, DeepSeek, Kimi, Qwen, and more — plus a $1.99 promo via npm i -g cline. Pricing, quota gaps, vs BYOK, and who should switch.
Jun 18, 2026
Tibo Sottiaux confirmed Codex runs any OSS model—not just GPT. This hands-on guide covers ollama launch codex, --oss flags, profile v2 config, model picks, 64k context requirements, and what still needs GPT (browser, computer use).
Jun 16, 2026
Ahmad (@TheAhmadOsman) shared VibeThinker 3B on June 16, 2026 — a model that reportedly matches Opus 4.5 (non-nerfed) at just 3B parameters. Built on Qwen 2.5-Coder with RL-based post-training. If the claims hold, this is the milestone that puts frontier-quality coding AI on a single consumer GPU.
Jun 30, 2026
InternScience's Agents-A1 claims trillion-class agent performance at 35B MoE scale — BrowseComp 75.5, GAIA 96.0, IFEval 94.8 — with open weights on Hugging Face and ModelScope. Here is what the paper claims, how to serve it, and the honest gaps on coding and Qwen lineage.
June 29, 2026 — 9:00 PM PT: Cline — the open-source coding agent behind the npm i -g cline install path — announced a $9.99/month subscription that bundles 2–5× discounted access to GLM-5.2 and other open-weight models: DeepSeek, Kimi, MiniMax, Mimo, and Qwen. A $1.99 promotional price applies when signing up through the npm flow shown in their launch video.
The thread hit 764K+ views in under a day. The timing lands squarely in the post–Fable 5 ban window: developers who lost unrestricted frontier API access are looking for managed open-weight routes that do not require juggling five provider dashboards every week.
| Question | Answer |
|---|---|
| Price | $9.99/mo standard; $1.99 promo via npm i -g cline sign-up flow |
| Discount claim | 2–5× cheaper vs buying the same models direct — volume-negotiated |
| Headline model | GLM-5.2 — Cline says they have been "impressed" with it |
| Also bundled | DeepSeek, Kimi, MiniMax, Mimo, Qwen |
| Where it runs | Cline CLI and Cline IDE (VS Code extension) |
| Exact quota | Not published — Cline says numbers shift as volume discounts improve |
| Best for | Devs tired of API key juggling who want one sub for Chinese open weights |
| Skip if | You need transparent token math, credits outside Cline, or only BYOK |
From the June 29 launch post:
We've been impressed with GLM-5.2 and so are introducing a $9.99/month subscription to give you 2-5x discounted access to it and other open weight models like DeepSeek, Kimi, MiniMax, Mimo, Qwen.
Use it on Cline CLI & IDE with $1.99 special promo if sign up via: npm i -g cline
Three structural bets are visible in that sentence:
That is reseller logic: Cline aggregates demand, negotiates with upstream APIs, and passes part of the savings as 2–5× effective discount. The trade-off is opacity — you buy convenience, not a published tokens-per-dollar table.
Since Fable 5 went offline globally on June 12, the default agent stack for many teams fractured. Some moved to BYOK with OpenRouter, others to GLM-5.2 harness guides, others to Kimi K2.7 or Qwen 3.7 weights.
Each path means another API key, another billing portal, and another model availability surprise when a provider throttles or re-prices. Cline's pitch addresses that operational tax directly.
Developers in the thread highlighted GLM-5.2 quality inside Cline, especially frontend work — consistent with BridgeBench coding strength reports, even where GLM remains text-only (vision is still a GLM-5.3 wishlist item).
The recurring praise theme: stop chasing providers every week. That matches how Kilo Code markets BYOK freedom — but Cline is betting many users prefer one curated bundle over 500-model choice paralysis.
@prathamdby asked for transparency — currently on OpenCode Go and might switch if usage limits are clear.
Cline's reply: "it's hard to say exact numbers since we actively trying to increase quota usage through volume based discounts!"
@quantumaidev pushed on the 2–5× discount definition — tokens or requests?
Cline's reply: "hard to say but we are actively trying to reduce costs and increase quota through increased volume"
explainx.ai read: Treat advertised discount as effective $/token vs retail API, not a fixed monthly cap. If you need hard budget ceilings for production teams, BYOK + provider dashboards or DeepSeek's published peak/off-peak tables remain more auditable.
@janekm reported following the sign-up flow from the video but not seeing $1.99 at checkout.
No public clarification followed in the captured thread. Before subscribing:
npm i -g cline
Then walk Cline's in-app billing and confirm the price line — promo eligibility may depend on account age, region, or a limited window Cline has not documented in the tweet.
| Option | Entry price | Model scope | Transparency |
|---|---|---|---|
| Cline open-weights sub | $9.99/mo ($1.99 promo) | GLM-5.2 + named Chinese open weights | Low — no public token table |
| Kilo Pass Starter | $19/mo | 500+ models via managed credits | Medium — credit tiers documented |
| BYOK (Cline / Kilo / OpenCode) | Pay providers directly | Full catalog you configure | High — provider invoices |
| OpenCode Go | Varies | Managed OpenCode subscription | Depends on OpenCode tier |
Cline occupies the cheapest managed bundle slot if the promo holds — but Kilo's OpenCode core and broader model catalog may win for teams already standardized on that stack.
Cline is the original open-source VS Code agent in a fork tree that shaped much of the ecosystem:
Cline → Roo Code → Kilo Code (OpenCode core)
If you already run Kilo Code with BYOK, Cline's subscription is additive, not a migration — unless you specifically want managed GLM-5.2 routing without a Z.ai API key.
For GLM-5.2 wiring across Pi, ZCode, Claude Code, and OpenCode without a Cline sub, see How to Run GLM 5.2 with Every Agent Harness.
Cline's launch ties billing to the standard install:
npm i -g cline
After install, open Cline in VS Code or run the CLI, then navigate to subscription / billing in product settings. Verify:
For MCP-heavy workflows, remember Cline supports MCP servers — skills and tools layer on top of whichever model the subscription routes.
Cline's announcement sits next to other June 2026 moves toward cheap managed open inference:
The through-line: frontier closed APIs get gated; open-weight stacks get cheaper and easier to consume. Cline's $9.99 bundle is an agent-host subscription on that curve — not a new model, but a distribution and billing layer that matters if key juggling was blocking adoption.
Official: Cline on X · npm: cline
Pricing, promo eligibility, and bundled models reflect Cline's June 29, 2026 announcement. Quota details may change as Cline negotiates volume discounts — verify in-product before budgeting. Last updated: July 1, 2026.