docs(skills): model prices from the live catalog, not a hand-kept table

skills/gentech-blockrun/SKILL.md

@@ -209,14 +209,16 @@ blockrun_video({ prompt: "animated data visualization", duration_seconds: 8 })
 | `blockrun_chat` (free mode) | $0 | NVIDIA-backed chat |
 | `blockrun_chat` (glm mode) | per-token | Zhipu GLM-5 coding — billed on tokens used, not a flat rate |
 | `blockrun_exa` (search) | $0.0110 | deep research (`contents`: $0.002/URL + $0.001) |
-| `blockrun_speech` | $0.0535/1k chars | TTS |
-| `blockrun_image` | $0.01675–0.106 | image generation |
-| `blockrun_music` | $0.1585 | music tracks |
-| `blockrun_video` | $0.421+ | short video clips (per-second × duration) |
+| `blockrun_speech` | per model — live `pricing` at `/api/v1/models` | TTS |
+| `blockrun_image` | per model — live `pricing` at `/api/v1/models` | image generation |
+| `blockrun_music` | per model — live `pricing` at `/api/v1/models` | music tracks |
+| `blockrun_video` | per model — live `pricing` at `/api/v1/models` | short video clips (per-second × duration) |
 | `blockrun_markets` | $0.0085 | prediction market data, wallet analytics (flat) |
 | `blockrun_modal` | $0.0020–$192 | remote sandbox execution — `timeout` drives the price, see the modal skill |
 | `blockrun_phone` | $0.0110–$5.001 | lookups; number buy/renew is $5.001 |
 
+Model-priced tools (chat beyond the free tier, speech, image, music, video) carry no number here because their prices move: read the model's `pricing` object from `GET https://blockrun.ai/api/v1/models` (Solana: `https://sol.blockrun.ai/api/v1/models`) — `input`/`output` per 1M tokens for chat, `per_1k_chars` for speech, `per_image`, `per_track`, or `per_second` for media, each plus `transaction_fee`. The HTTP 402 challenge quotes the exact amount before anything is signed.
+
 ### Daily Budget Formula (GenTech's Setup)
 
 ```yaml

skills/image-prompting/SKILL.md

@@ -22,23 +22,23 @@ Most image failures are prompt failures. This skill gives the MCP agent a repeat
 
 ## Quick Decision Table
 
-Costs are what you are actually CHARGED, verified live. **Size is the biggest
+**Prices:** per image, and they change — look the model up live rather than trusting a number here: `GET https://blockrun.ai/api/v1/models` (Solana: `https://sol.blockrun.ai/api/v1/models`), find the entry whose `id` is the model below, and read `pricing.per_image` plus `pricing.transaction_fee`. That is the base-size rate; the exact amount for your size and quality is quoted in the HTTP 402 challenge before anything is signed. **Size is the biggest
 lever:** any dimension above 1024 moves GPT Image 2 to the large tier and roughly
 doubles the price ($0.064 → $0.127). Ask for `1536x1024` only when you need it.
 
-| User wants... | Model | Mode | Size | Cost |
-|---|---|---|---|---|
-| Poster / typography-heavy asset | `openai/gpt-image-2` | generate | `1536x1024` or `1024x1536` | $0.127 |
-| Clean product / UI mockup | `openai/gpt-image-2` | generate | `1024x1024` | $0.064 |
-| Photoreal / fashion / editorial | `openai/gpt-image-2` or `google/nano-banana-pro` | generate | `1024x1024` | $0.064–0.106 |
-| Pro-level photoreal at Flash speed | `google/nano-banana-2` | generate | `1024x1024` (only size) | $0.0955 |
-| Artistic / stylized / fast | `google/nano-banana` | generate | `1024x1024` | $0.0535 |
-| Cheapest usable draft | `zai/cogview-4` | generate | `1024x1024` | $0.01675 |
-| Widescreen / banner on a budget | `bytedance/seedream-5-pro` | generate | `2048x1024` or `1280x720` | $0.04825 ($0.0955 when both sides >1024) |
-| Premium render or high-precision edit | `openai/gpt-image-2.5-sunburst` (`-flare` generates only, same price) | generate or edit | `1024x1024` | $0.295 ($0.589 at `1536x1024` / `1024x1536`) |
-| Stylized Grok render, one step above the base | `xai/grok-imagine-image-2.0` | generate | `1024x1024` (only size) | $0.043 |
-| Edit an existing image (localized change) | `openai/gpt-image-2` | edit | match source | $0.064 at 1024x1024, $0.127 above |
-| Composite from multiple refs | `openai/gpt-image-2` | edit (multi-ref) | match target | $0.064 at 1024x1024, $0.127 above |
+| User wants... | Model | Mode | Size |
+|---|---|---|---|
+| Poster / typography-heavy asset | `openai/gpt-image-2` | generate | `1536x1024` or `1024x1536` |
+| Clean product / UI mockup | `openai/gpt-image-2` | generate | `1024x1024` |
+| Photoreal / fashion / editorial | `openai/gpt-image-2` or `google/nano-banana-pro` | generate | `1024x1024` |
+| Pro-level photoreal at Flash speed | `google/nano-banana-2` | generate | `1024x1024` (only size) |
+| Artistic / stylized / fast | `google/nano-banana` | generate | `1024x1024` |
+| Cheapest usable draft | `zai/cogview-4` | generate | `1024x1024` |
+| Widescreen / banner on a budget | `bytedance/seedream-5-pro` | generate | `2048x1024` or `1280x720` |
+| Premium render or high-precision edit | `openai/gpt-image-2.5-sunburst` (`-flare` generates only, same price) | generate or edit | `1024x1024` |
+| Stylized Grok render, one step above the base | `xai/grok-imagine-image-2.0` | generate | `1024x1024` (only size) |
+| Edit an existing image (localized change) | `openai/gpt-image-2` | edit | match source |
+| Composite from multiple refs | `openai/gpt-image-2` | edit (multi-ref) | match target |
 
 **Valid GPT Image 2 sizes:** `1024x1024` (square), `1536x1024` (landscape ~3:2), `1024x1536` (portrait ~2:3).