Model & API Radar
Model catalogs and free-API terms were checked on 2026-09-12. I would recheck both the public docs and account limits before integrating.
After reading through material from seven vendors, the biggest change is not another claim to the “best model.” It is that product lines finally make more sense. Flagships handle the hard cases, mid-tier models carry routine volume, and small models run repetitive work cheaply. Image, video, and realtime voice are also moving toward dedicated endpoints.
So I no longer start with “which company is best?” I start with three questions: does this job deserve a flagship, can a mid-tier model handle it, and is the extra control of open weights worth operating the stack myself?
Hosted APIs: choose the tier before the vendor
The legacy deepseek-v4-flash and deepseek-v4-flash-vision-exp IDs remain accepted but now route to V4.1 Flash and its pricing; they no longer select the retired models.
In practice I would start in the middle tier. Terra, Sonnet, Gemini Flash, Grok, and DeepSeek V4.1 Flash / V4 Pro can all enter the first test. I would test a higher-capability tier such as Astra, Sol, Fable 5.1, or Opus only after repeated task failures justify its cost. In the other direction, extraction, classification, and format conversion that already work reliably should keep moving toward Luna, Haiku, or another cheaper tier.
That is cheaper than throwing a flagship at every problem and less frustrating than optimizing for the lowest token price from day one.
Free APIs: keep a shortlist, not a permanent leaderboard
For a personal prototype, I now keep three main routes: Gemini for long context and multimodal work, Groq for low latency, and OpenRouter for trying several free models behind one interface. All can start at zero payment, but they differ sharply in units, reset windows, and data terms.
The most useful update is not another provider joining the list, but two leaving it. Cerebras now offers a 5 USD, 30-day Free Trial after a verified payment method, not an automatically renewing free tier. GitHub Models was fully retired on 2026-07-30. Even OpenRouter's own June comparison was already stale on both points by August.
I therefore would not build around one free endpoint. A personal project can start with OpenRouter for interface consistency, then keep direct Gemini and Groq integrations for clearly assigned workloads. Cloudflare or GLM only joins when its ecosystem or language advantage matters. Every route still needs 429 handling, Retry-After, a daily budget, and failover; none should receive private code or customer data by default.
This section is about API inference allowance, not usage bundled into terminal agents. That separate accounting problem lives in Free Allowances for AI Coding CLIs.
Local inference is a separate decision
When data locality or stable volume justifies running a service, move to the local AI stack. Hardware fit, coding candidates, runtimes, and evaluation each have a separate owner there; this hosted radar does not duplicate their snapshots.
Prices deserve one glance, not a strategy
These are catalog rates checked on 2026-09-12, in USD per million input/output tokens for standard text usage, not total task costs or account entitlements. Caching, batch jobs, reasoning, long context, and contracts can change the actual bill substantially.
I would not switch models from this table alone. A call that costs half as much can still be more expensive if it needs retries, produces repairs, or keeps failing at tool use. The useful number is the cost of finishing one real task, not the sticker price of one token.
If I were starting a project today
My order would be:
- run five real tasks with the mid-tier model that fits the existing workflow best;
- keep the failures and separate model problems from tool or prompt problems;
- move only the stubborn tasks to a flagship;
- try cheaper models on work that has become repetitive and predictable;
- move to the local AI stack when data must stay local or the workload justifies operating a service.
The vendor material is enough to shrink the shortlist, but it is not a shared benchmark. I am not going to crown an overall winner from product pages. The useful next step is to replay the same work on my own repositories and documents, with failure criteria that matter to me. Only then is a default model actually personal.