Qwen3.5 4B

qwen/qwen3.5-4b

Qwen (Alibaba)· 2026-03-02· 262K ctx· apache-2.0

open weightstool callreasoning quant: unknown

spread (output, p90 ÷ p10)

— too few providers

$1.00 – $1.00 · 1 providers · max÷min 1.0×

1 provider priced · 2 offerings

Including routers & resellers

Cheapest / median / spread use providers that serve the model on their own capacity: first-party lab, hyperscaler, inference host.

Provider classes in the stats:

Blended $/M assumptionsInput : output: 1Cache-hit share of input%

blended = 90.9% input × [(1 − 70% cache) × input + 70% cache × cache-read] + 9.1% × output + cache writes (30% of uncached input) where a write price is published

A provider that publishes no cache price is blended with cache share 0 (all input at the input price) and labelled “no cache price published” — never penalised against identical in/out prices.

Provider	Input	Output	Cache rd	Blended
EmpirioLabs AI unclassified	$0.0400	$0.0700	$0.0200	$0.0300
Merge Gateway router / aggregator	$0.0400	$0.0700	—	$0.0427no cache price published
Thinking Machines first-party lab	$0.33	$1.00	$0.07	$0.22

Unpriced · 2

Listed at $0 with no published pricing — we can't tell whether it's free, bundled or just missing. Excluded from cheapest, median and spread.

QVAC qwen3.5-4bunpriced
SiliconFlow (China) Qwen/Qwen3.5-4BMainland-China endpointunpriced

Connect

Provider (default: cheapest output among direct providers)

$1.00 /M · first-party lab

opencode.json

{
  "provider": {
    "thinkingmachines": {
      "npm": "@ai-sdk/anthropic",
      "options": {
        "baseURL": "https://tinker.thinkingmachines.dev/services/tinker-prod/anthropic/api/v1"
      },
      "models": {
        "Qwen/Qwen3.5-4B": {}
      }
    }
  }
}

Set TINKER_API_KEY in your environment. Generated — verify against the OpenCode docs.

Not enough history yet — the min/median/max sparkline fills in as the daily snapshot accumulates.

Data: models.dev (MIT), snapshot 2026-10-21. Prices are vendor-published; verify before purchase. Methodology.