pi-freeflow 1.3.8 → 1.3.9

package/README.md

@@ -23,14 +23,15 @@ You love Pi. You love free models. You hate:
 - `reasoning_effort: max` rejected and thinking silently disabled
 - Paying for 1M context you should get for free
 
-**FreeFlow fixes all three — in 17 files, 102KB.**
+**FreeFlow fixes all three — in 17 files, 115KB.**
 
 ### Features
 
 | What You Get | Value | Cost |
 | :--- | :--- | :--- |
-| **23 Curated Free Models** — 9 OpenCode Zen + 14 Kilo, 66K-1M context, vision-ready | Unlimited | **$0** |
+| **23 Curated Free Models** — 9 OpenCode Zen + 14 Kilo Gateway, 66K-1M context, vision-ready | Unlimited | **$0** |
 | **BYO Relay Pool** — Cloudflare Workers + Vercel Edge, round-robin, seamless `429 → next` | Auto-scale | **$0** (your free tiers) |
+| **Adaptive Health & Error Detection** — Auto-cooldown on 429/504/socket drops, instant roll to healthy relays | Self-healing | **$0** |
 | **Dumb Proxy That Never Breaks** — `127.0.0.1:18080`, host-normalized, pathname-guarded `/v1/models` | 100% uptime | **$0** |
 | **Real Logs** — `~/.pi/agent/pi-freeflow.log`, 5MB rotate, debug toggle | Observable | **$0** |
 
@@ -45,26 +46,41 @@ Philosophy: **Thin by design.** We only ship model list + relay proxy + log. Hos
 ```
 
 #### OpenCode Zen (9) — Responses & Chat API
-Best for long-context reasoning. All vision-ready models marked.
-
-| Model | Context | Max Output | Thinking | Vision |
-| :--- | :--- | :--- | :--- | :--- |
-| `claude-sonnet-4.5-free` / `muse-spark-1.2-contributor-free` | 1M | 131K | `minimal → max` | ✅ |
-| `deepseek-v4-flash-free` | 1M | 384K | `minimal → max` | ❌ |
-| `grok-code-fast-1-preview-f-free` / `x-preview-f-free` | 1M | 131K | `low / high / max` | ✅ |
-| `mimo-v2.5-free` | 1M | 131K | `low / medium / high` | ✅ |
-| `minimax-m2.1-free` / `laguna-s-2.1-free` | 1M | 131K | `minimal → max` | ❌ |
-| `nemotron-3.5-lightning-free` | 1M | 262K | `minimal → max` | ❌ |
-| `nemotron-3-ultra-free` | 1M | 128K | `minimal → max` | ❌ |
-| `qwen3-coder-480b-free` / `hy3-free` | 262K | 128K | `minimal → max` | ❌ |
-| `big-pickle` | 200K | 32K | `high / max` | ❌ |
+Best for long-context reasoning & autonomous agentic tasks.
+
+| Model ID | Creator / Lab | Context | Max Output | Thinking | Vision |
+| :--- | :--- | :--- | :--- | :--- | :--- |
+| `muse-spark-1.2-contributor-free` | Meta Superintelligence Labs | 1M | 131K | `minimal → max` | ✅ |
+| `x-preview-f-free` | Ox Alpha | 1M | 131K | `low / high / max` | ✅ |
+| `mimo-v2.5-free` | Xiaomi MiMo | 1M | 131K | `low / medium / high` | ✅ |
+| `deepseek-v4-flash-free` | DeepSeek | 1M | 384K | `minimal → max` | ❌ |
+| `laguna-s-2.1-free` | Poolside | 1M | 131K | `minimal → max` | ❌ |
+| `nemotron-3.5-lightning-free` | NVIDIA | 1M | 262K | `minimal → max` | ❌ |
+| `nemotron-3-ultra-free` | NVIDIA | 1M | 128K | `minimal → max` | ❌ |
+| `hy3-free` | Tencent Hunyuan | 262K | 262K | `minimal → max` | ❌ |
+| `big-pickle` | Big Pickle | 200K | 32K | `high / max` | ❌ |
 
 #### Kilo Gateway (14) — OpenRouter Compatible
-Zero-config, `Bearer kilo-free`.
+Keyless access with `Bearer kilo-free`. Short slash-free & colon-free CLI aliases supported.
+
+| Model ID | Creator / Lab | Context | Max Output | Thinking | Vision |
+| :--- | :--- | :--- | :--- | :--- | :--- |
+| `dots-3-note-preview` (`dots-studio/...:free`) | Dots Studio | 512K | 512K | `openrouter` (auto) | ✅ |
+| `step-3.7-flash` (`stepfun/...:free`) | StepFun | 262K | 262K | `low / medium / high` | ✅ |
+| `nemotron-3-nano-omni` (`nvidia/...:free`) | NVIDIA | 256K | 65K | `openrouter` (auto) | ✅ |
+| `nemotron-3-ultra-550b` (`nvidia/...:free`) | NVIDIA | 1M | 65K | `openrouter` (auto) | ❌ |
+| `nemotron-3.5-lightning` (`nvidia/...:free`) | NVIDIA | 1M | 131K | `openrouter` (auto) | ❌ |
+| `nemotron-3-super` (`nvidia/...:free`) | NVIDIA | 262K | 262K | `openrouter` (auto) | ❌ |
+| `hy3` (`tencent/hy3:free`) | Tencent Hunyuan | 262K | 262K | `openrouter` (auto) | ❌ |
+| `north-mini-code` (`cohere/...:free`) | Cohere | 256K | 64K | `openrouter` (auto) | ❌ |
+| `laguna-s-2.1` (`poolside/...:free`) | Poolside | 1M | 131K | `openrouter` (auto) | ❌ |
+| `laguna-xs-2.1` (`poolside/...:free`) | Poolside | 262K | 32K | `openrouter` (auto) | ❌ |
+| `lfm-2.5` (`liquid/lfm-2.5-2.6b:free`) | Liquid AI | 128K | 32K | `openrouter` (auto) | ❌ |
+| `kilo-auto` (`kilo-auto/free`) | Kilo Gateway Auto | 256K | 10K | ❌ | ❌ |
+| `openrouter` (`openrouter/free`) | OpenRouter Free | 200K | 65K | ❌ | ✅ |
+| `content-safety` (`nvidia/...:free`) | NVIDIA | 128K | 8K | ❌ | ✅ |
 
-`dots-3-note-preview:free` (512K), `step-3.7-flash:free` (262K), `nemotron-3-nano-omni:free` (256K), `nemotron-3-ultra-550b:free` (1M), `nemotron-3-super:free` (262K), `hy3:free` (262K), `north-mini-code:free` (256K), `minimax-m2.1/xs:free` (262K), `lfm-2.5:free` (66K), `kilo-auto/free`, `openrouter/free`, `content-safety:free` + 2 more.
-
-> All `off → null` (hidden). Host decides token budgets, not us.
+---
 
 ### How It Works — The Magic: BYO Relays, Zero Rate Limits
 
@@ -74,10 +90,11 @@ You → 127.0.0.1:18080 (dumb proxy, host-normalized) → x-relay-target → N e
 ```
 
 - **Round-robin per request** — 10 parallel subagents hit N different egress IPs (your pool size). No thundering herd.
+- **Adaptive Health & Error Cooldown** — Relays hitting 429, 504, or network socket drops enter automatic cooldown (30-90s) and move behind healthy candidates.
 - **Seamless 429 roll** — `429 / 502 / 503 / 504` → instant roll to next relay, never 429 to agent.
 - **Exhausted? Direct fallback** — If all relays are 429 (global limit), transparent direct fetch. Host shows rate limit only when truly global.
 - **No live catalog fetch** — 24h `DISK_CACHE_ONLY`, never fetch from subagent → no `Unable to connect`.
-- **Real log (v1.3.1 thin):** `candidates:N → relay-A 429 rolling → relay-B 200 in 1.2s → 98 chunks 663KB done`
+- **Stream Truncation Resilience** — Detects SSE terminal events; handles premature stream closure with clean termination.
 
 You bring the relays (free tiers). We bring the rolling.
 
@@ -103,10 +120,10 @@ pi install npm:pi-freeflow
 
 ```bash
 omp
-/model → freeflow → claude-sonnet-4.5 (1M) → max
+/model → freeflow → muse-spark-1.2-contributor-free (1M) → max
 
 # or CLI
-omp -p --model freeflow/claude-sonnet-4.5-contributor-free "build me a SaaS"
+omp -p --model freeflow/muse-spark-1.2-contributor-free "build me a SaaS"
 ```
 
 #### 3. (Optional but 🔥) Add Your Free Relays
@@ -131,6 +148,7 @@ omp → /freeflow use https://your.workers.dev
 **Verify your pool:**
 ```bash
 /freeflow status        # relay-A 1/N (ON) → candidates:N
+/freeflow list          # lists all relays with health status
 /freeflow logs          # tail -25
 cat ~/.pi/agent/pi-freeflow.log | tail -n 20
 ```
@@ -158,7 +176,7 @@ Log rotation at 5MB. Clean, parseable, real-time.
 | 400 loops N× relays | Narrow 429/5xx only, 400 surfaces instantly |
 | `GET /v1/models?foo` leaks paid | Pathname-guarded |
 
-**Result: <500 LOC core, 29 tests PASS, typecheck PASS.**
+**Result: <500 LOC core, 39 tests PASS, typecheck PASS.**
 
 ### FAQ
 
@@ -195,5 +213,5 @@ MIT © trefeon
 
 Start now:
 ```bash
-omp plugin install pi-freeflow && omp --model freeflow/claude-sonnet-4.5-contributor-free --thinking max "make it sell"
+omp plugin install pi-freeflow && omp --model freeflow/muse-spark-1.2-contributor-free --thinking max "make it sell"
 ```

package/package.json

@@ -1,7 +1,7 @@
 {
 	"name": "pi-freeflow",
 	"type": "module",
-	"version": "1.3.8",
+	"version": "1.3.9",
 	"description": "Thin provider for OMP/Pi — model list + dumb relay proxy + log; host pi-ai owns thinking/normalization",
 	"main": "extensions/index.ts",
 	"types": "src/index.ts",

package/src/models.ts

@@ -87,7 +87,7 @@ export const OPENCODE_MODELS: ModelDef[] = [
 		name: "Hy3 (262K)",
 		reasoning: true,
 		contextWindow: 262_144,
-		maxTokens: 128_000,
+		maxTokens: 262_144,
 		input: ["text"],
 		thinkingLevelMap: {
 			off: null,
@@ -220,7 +220,7 @@ export const KILO_MODELS: ModelDef[] = [
 		name: "Nemotron 3.5 Lightning (Kilo)",
 		reasoning: true,
 		contextWindow: 1_000_000,
-		maxTokens: 65_536,
+		maxTokens: 131_072,
 		input: ["text"],
 		thinkingFormat: "openrouter",
 	},
@@ -238,9 +238,8 @@ export const KILO_MODELS: ModelDef[] = [
 		name: "Tencent Hy3 (Kilo)",
 		reasoning: true,
 		contextWindow: 262_144,
-		maxTokens: 128_000,
+		maxTokens: 262_144,
 		input: ["text"],
-		thinkingFormat: "openrouter",
 	},
 	{
 		id: "cohere/north-mini-code:free",
@@ -255,8 +254,8 @@ export const KILO_MODELS: ModelDef[] = [
 		id: "poolside/laguna-s-2.1:free",
 		name: "Laguna S 2.1 (Kilo)",
 		reasoning: true,
-		contextWindow: 262_144,
-		maxTokens: 32_768,
+		contextWindow: 1_048_576,
+		maxTokens: 131_072,
 		input: ["text"],
 		thinkingFormat: "openrouter",
 	},
@@ -273,8 +272,8 @@ export const KILO_MODELS: ModelDef[] = [
 		id: "liquid/lfm-2.5-2.6b:free",
 		name: "Liquid LFM 2.5",
 		reasoning: true,
-		contextWindow: 65_536,
-		maxTokens: 8_192,
+		contextWindow: 128_000,
+		maxTokens: 32_768,
 		input: ["text"],
 		thinkingFormat: "openrouter",
 	},
@@ -292,7 +291,7 @@ export const KILO_MODELS: ModelDef[] = [
 		reasoning: false,
 		contextWindow: 200_000,
 		maxTokens: 65_536,
-		input: ["text"],
+		input: ["text", "image"],
 	},
 	{
 		id: "nvidia/nemotron-3.5-content-safety:free",
@@ -300,7 +299,7 @@ export const KILO_MODELS: ModelDef[] = [
 		reasoning: false,
 		contextWindow: 128_000,
 		maxTokens: 8_192,
-		input: ["text"],
+		input: ["text", "image"],
 	},
 ];